Datasets:

gabrielaltay

/

pmcoa

like 0

==== Front Lancet Lancet Lancet (London, England) 0140-6736 1474-547X Elsevier Ltd. S0140-6736(21)01154-5 10.1016/S0140-6736(21)01154-5 Perspectives Cultures of contagion Prasad Aarathi 29 5 2021 29 May-4 June 2021 29 5 2021 397 10289 20412041 © 2021 Elsevier Ltd. All rights reserved. 2021 Elsevier Ltd Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. ==== Body pmcThe Science Gallery Bengaluru's online exhibition CONTAGION opened in April, 2021, just as India approached a staggering 20 million reported cases of COVID-19, a number likely to be a vast undercount. The alarming second wave of COVID-19 in India gives this exhibition an immediacy and relevance. Although we live with the pandemic still, there is a place for contextualising our experiences of COVID-19 through recent global history. There is a role for fresh ways of engendering discussion as this pandemic runs its course and the tide of so many personal and collective griefs surges. How could transmission on this scale be both so predictable and have taken us by surprise? For Jahnavi Phalkey, the exhibition's co-curator and Founding Director of Science Gallery Bengaluru, a public institution for research-based engagement, the idea of CONTAGION came long before the current crisis in India. When it took shape, it became an expansive examination of the spread not only of diseases, but also of the transmission of behaviours, ideas, and emotions in culture and society. “We are all differently inspired, motivated, and stimulated; our preference could be aural, visual, textual, or haptic, among others”, Phalkey says: “similarly, we respond differently at different times in our lives to conditions around us. We wanted to make this exhibition season approachable from these different perspectives, motivations, and desires, which would explain, to some extent, why we wanted to broaden the idea of contagion to nestle comfortably with transmission beyond disease.” A historian of science, Phalkey says the exhibition “aims to bridge the distance between research and the public at large”, an objective that the Science Gallery Bengaluru hopes will help the exhibition's audiences “develop a critical appreciation of knowledge and perhaps, eventually, raise better questions in public debate”. CONTAGION features 12 diverse exhibits. A Cluster of 17 Cases, made by the artist collective Blast Theory uses interactive media to virtually enter Hong Kong's Metropole Hotel on the fateful night a man attending a family wedding infected 16 hotel guests who, the exhibit shows, would play a part in spreading SARS across the world. In Mapping Cholera, journalist Sonia Shah and data artist Dan McCarey brilliantly use interactive maps to render visible the spread, magnitude, and scale of two cholera outbreaks almost two centuries apart: one in New York, USA, in 1832, two decades before John Snow's famous mapping of cholera in London, UK; the other brought by UN peacekeepers to Haiti in 2010, in the wake of the earthquake that had already devastated the country. Another collection, featuring work by social anthropologist Christos Lynteris, forms a compelling visual history of the bubonic plague that arrived in Mumbai in 1896, the first such disease to be extensively documented through photography in India. Called simply, Controlling the Plague in British India, the faded black and white and sepia tones that show the burning or disinfecting of contaminated places or encampments to which populations were removed are difficult to look away from, particularly at a time when so many people in India are living through an escalating COVID-19 crisis and with the knowledge that, by its end, that plague killed an estimated 12 million people globally. Allied to these exhibits are a series of hugely relevant videos and lectures that frame the art with indepth analyses of topics, from the management of COVID-19 in India and health and justice in a pandemic, to community care, conspiracies, and trust. Contrasting narratives are presented in CONTAGION: death, stigma, the movement of microbes, and the inner lives of viruses all feature. But there are also intriguing ideas that come from unexpected sources: a live stream of an ant colony in which its residents make chemicals that might help to tackle antimicrobial resistance; a museum in which computer malware transmits creative, rather than malicious, messages; and animations that make sense of the digital world as information and misinformation multiplies. By capturing the morbid alongside the playful, CONTAGION succeeds in what Phalkey's team set out to do. “A terrifying pandemic could be bracketed out, ignored, or engaged with”, she says. “We wanted to make the journey to understanding this phenomenon less terrifying, more engaging, and at times, interesting enough to provide some relief from the relentless suffering surrounding the global collective experience of the pandemic.” © 2021 Controlling the Plague in British India/“Flushing engine cleaning infected houses”, Moss C, Captain, fl. ca. 1897/The Bombay plague epidemic of 1896–1897: work of the Bombay Plague Committee/Photographs attributed to Capt C Moss, 1897, Wellcome Collection 2021 CONTAGION Online exhibition until June 13, 2021 Science Gallery Bengaluru, National Centre for Biological Sciences (Tata Institute of Fundamental Research), Bengaluru, Karnataka, India https://bengaluru.sciencegallery.com/contagion

0

PMC9753768

NO-CC CODE

2022-12-16 23:26:18

no

Lancet. 2021 May 29 29 May-4 June; 397(10289):2041

utf-8

Lancet

10.1016/S0140-6736(21)01154-5

oa_other

==== Front Lancet Lancet Lancet (London, England) 0140-6736 1474-547X Elsevier Ltd. S0140-6736(21)01647-0 10.1016/S0140-6736(21)01647-0 Perspectives Past and present women pioneers in biomedical science Marks Lara a a Jeffrey Cheah Biomedical Centre, University of Cambridge, Cambridge CB2 0AW, UK 22 7 2021 24-30 July 2021 22 7 2021 398 10297 293295 © 2021 Elsevier Ltd. All rights reserved. 2021 Elsevier Ltd Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. ==== Body pmcCOVID-19 has propelled a number of scientific breakthroughs that have only been possible because of unprecedented global research collaborations. These remarkable achievements will have profound implications for the future. Women have been front and centre in many of these developments, notably in the area of COVID-19 vaccines and sequencing of SARS-CoV-2. Yet in this pandemic women's position in biomedicine has not been helped by the fact that many women have been disproportionately affected by increased caring responsibilities during lockdowns, reducing the amount of time they have been able to devote to research and publication—with implications for career advancement. As a historian, I have spent many years uncovering the contributions of women to advances in medicine and science, many of whose achievements were overshadowed by their male counterparts. So it is illuminating to mark the scientific achievements of a few of the women involved in the COVID-19 response and also look back to the work of other forgotten female researchers in whose footsteps they tread. All too often women get ignored because their skills are less highly prized and their discoveries and input are marginalised within the workplace. Historically, women have tended to land up in medical and scientific fields that were traditionally deemed less prestigious but where they had greater freedom to work part-time. Two such areas are vaccines and microbiology. From the 1970s there was a general decline in pharmaceutical companies' involvement in vaccine research, partly connected to an increase in lawsuits against vaccine manufacturers and paucity of government investment. Similarly, clinical microbiology had suffered years of underfunding. In 2003, the UK Government cut funding for the Public Health Laboratory Service, which since 1946 had been one of the country's first lines of defence against public health threats. Over the past three decades, clinical microbiology research in the USA has also been underfunded. One reason for the limited investment in vaccines and microbiology in previous decades was a mistaken belief that infectious diseases no longer posed a threat to public health in high-income countries. Yet both vaccines and microbiology have proven crucial in COVID-19 control. It is in these fields where women have made a visible mark in the pandemic, often after working for many years below the radar. One such woman is Sarah Gilbert, professor of vaccinology at the Jenner Institute, Oxford University, UK, who, together with colleagues, designed the platform that underpins the AstraZeneca–Oxford ChAdOx1 nCoV-19 vaccine. For more than 20 years, Gilbert has been working with Adrian Hill, director of the Jenner Institute, to create recombinant viral vectors that they believed would make it easier to make vaccines capable of inducing T-cell responses alongside antibody responses, which would enhance their efficacy. The ChAdOx1 nCoV-19 vaccine uses a chimpanzee adenovirus vector to deliver a genetic sequence that codes for the surface spike protein of SARS-CoV-2 to prime the immune system to destroy the coronavirus. Gilbert's research group first started genetically modifying a common cold virus in the 1990s. Patented in 2016, the chimpanzee-derived adenoviral vector provided a platform to rapidly roll out cost-effective vaccines against emerging pathogens and neglected diseases. Before COVID-19, Gilbert's team explored the vector's capability in a number of vaccines, including against influenza, malaria, and the Ebola and Zika viruses. Early stage human trials in 2018 indicated the vector to be safe and effective at promoting an immune response in a vaccine for MERS-CoV. Despite this impressive track record, like some other female scientists Gilbert once considered giving up her scientific career because of the challenges of balancing the demands of full-time research with her role as a parent to triplets. Yet, as she admits, parenthood prepared her well for coping with the pressures involved in her accelerated research during this pandemic. It taught her how to be highly organised and how to survive on very little sleep. As she puts it, “If you get 4 hours a night with triplets, you're doing very well. I've been through this.” Importantly, her research efforts were helped by the fact she and her team were already immersed in work on pandemic preparedness to create a vaccine against an outbreak of a new disease. Gilbert's grit is matched by that of Professor Katalin Karikó, whose once dismissed dream to develop mRNA therapeutics to fight disease lies behind the development of the Pfizer–BioNTech and Moderna COVID-19 vaccines. Born in Hungary and trained as a biochemist, Karikó has had more than her fair share of challenges. In 1985, she was made redundant at the age of 30 and forced to seek a new academic life with her husband and young daughter in the USA with little funds to support them. Working in Philadelphia, Karikó experienced 15 years of having grant applications rejected for research on therapeutic applications of mRNA, an area she first started investigating in Hungary. Without grant support for her research, in 1995 Karikó had the demoralising experience of being denied promotion at the University of Pennsylvania, although she had been on the faculty for 6 years. Karikó's struggle partly reflected the fact that her research was seen as too radical because synthetic mRNA can be easily degraded, which made it potentially unsuitable to be a drug. Unwilling to give up on her idea, Karikó's luck changed when, in 1997, she persuaded Drew Weissman, a physician and an immunologist who had just joined the faculty at her institution, to collaborate with her. In 2000 they reported synthetic mRNA to be immunogenic, but in 2005 made a breakthrough: they found that incorporating modified nucleosides into mRNA prevented the immune response. Hardly noticed at the time, their discovery laid a foundation for using mRNA technology for COVID-19 vaccines and Karikó is now a senior vice president at BioNTech. Fanny Angelina Hesse (1850–1934) © 2021 GL Archive/Alamy Stock Photo 2021 That technology played a part in the work of another notable researcher. Working persistently to advance her field, from 2014 viral immunologist Kizzmekia Corbett led research into the development of new vaccines for coronaviruses at the US National Institutes of Health (NIH) Vaccine Research Center. Corbett, who grew up in North Carolina, USA, was first drawn to coronaviruses because, as she says, “ironically…most other people were not” and it was a way to “build a niche and to tap into some uncharted territory”. Crucially, Corbett helped to determine that an ideal vaccine target for such viruses was its spike protein. She worked this out with colleagues in 2017 while investigating MERS-CoV. This research laid an important blueprint for the Moderna–NIH mRNA-1273 vaccine that Corbett helped to develop. Corbett had a key role in designing the vaccine and led the preclinical studies needed before it could be tested in humans. She also developed the assays needed for testing clinical trial samples. Corbett also contributed to work on antibody responses to coronaviruses that provided a foundation for the development of the neutralising monoclonal antibody therapy LY-CoV555 for COVID-19. One of the first generation in her family to attend university, Corbett first developed a curiosity for science aged 16 years, after being selected by a programme for gifted minority students to work as an intern in a chemistry laboratory at the University of North Carolina. Recently appointed Assistant Professor of Immunology and Infectious Diseases at Harvard University, Corbett has written about the importance to her of “‘Each one teach one’, an African American proverb, birthed out of slavery, suggesting it is one's duty to pass knowledge onward to those who are not as privileged.”' Eager to mentor other girls and minority students to follow in her footsteps, Corbett has made contributions in the USA to increasing participation in clinical trials and to the uptake of COVID-19 vaccines among Black communities that have been hit disproportionately by COVID-19. Just as Corbett was the first in her family to attend university, so too was Sharon Peacock. Professor of public health and microbiology at the University of Cambridge, in March, 2020, Peacock spearheaded the setting up of the COVID-19 Genomics UK Consortium (COG-UK) with colleagues to rapidly sequence as many SARS-CoV-2 viral genomes as possible to map out the spread and evolution of the virus. Discussing COG-UK with me, Peacock is astonished at the speed with which she and others managed to get the ball rolling, at a time when many were initially sceptical that the virus would mutate enough to make the venture worthwhile. Poignantly, she remembers occasions when she thought, “What have I done? What if this is a complete failure…and it's the wrong decision.” For Peacock “a key defining moment” in her life was when she failed the exam needed to gain entrance to grammar school. This meant she went to a secondary school where more emphasis was placed on practical domestic and secretarial skills than academic subjects, preventing her from taking exams necessary for entering university. Only equipped with a basic training in maths, Peacock left school aged 16 years to start work in a shop and then a year later set about training first as a dental nurse and then a nurse before deciding to pursue a career in medicine. She attained her entry qualifications by going to evening classes and a technical college while working part-time. Peacock eventually secured a place in medical school as a mature student. Before the pandemic, Peacock had spent a decade determining the value of pathogen sequencing in public health microbiology. This included the use of genomic epidemiology to understand the transmission of health-care-associated infections as well as the relationship between antimicrobial resistance (AMR) in humans, livestock, and the environment. Her experience in this area was a perfect background for COG-UK. Peacock attributes much of the consortium's success to the chemistry and “can do” attitude of the people she managed to get involved. It is also reflective of her own collaborative leadership style that encourages everyone to have an equal voice. From the start she envisioned the consortium should have a clear vision and structure that allowed independent thinkers to work loosely together, maximising on the possible impact of the varied expertise across the membership. Peacock's collaborative approach is refreshing in today's highly competitive world of science. It also brings to mind some other women from the past who I profiled in an online exhibition about scientists' long efforts to understand and overcome the challenge of AMR. What is striking is how much of the early knowledge generated about the biological mechanisms underpinning AMR rests on women's ingenuity, yet their contributions have been largely forgotten. The microbiology world, for example, owes a great deal to Fanny Angelina Hesse, known as Lina. It is thanks to her that laboratories now use agar to culture and study microbes. She first suggested the medium as a replacement for gelatin in 1881 while working as an unpaid technician to her husband, Walther Hesse, a medical practitioner and bacteriologist. For many years she coated glass tubes with gelatin for Walther to grow microorganisms, but this was unsatisfactory as the gelatin melted on warm days and could be destroyed by some bacteria. To solve the problem, Lina turned to agar, which she used to prepare preserves and puddings. Agar had several advantages: it remained solid in temperatures up to 90°C, was transparent, could not be digested by microbial enzymes, and could be sterilised and stored for long periods. This meant agar could be used for long-term cultures of bacteria, which was crucial, for example, in research on tuberculosis. A talented artist with a strong understanding of bacteriology and microscopy, she also drew many pictures of magnified colonies of bacteria at different growth phases for her husband's publications. Walther, however, rarely acknowledged her assistance. Lina herself never sought credit. Her contributions only came to light in 1992 when her grandson published a brief biography of Lina and Walther. Esther Lederberg is another woman largely overshadowed by her husband, Joshua Lederberg. Awarded the Nobel Prize in Physiology or Medicine in 1958 for his discovery that bacteria can exchange genes, Joshua's breakthrough rested considerably on Esther's input, to whom he gave scant acknowledgment when he received the prize. Esther was herself a pioneer in bacterial genetics but never gained a permanent academic position. In 1950 she isolated the lambda phage, a virus that infects Escherichia coli, and in 1953 published work showing that it could insert its own DNA into the genome of infected bacteria. The phage provided an ideal tool to understand the pathways that viruses use to transfer genes into bacterial cells, which is a key mechanism behind AMR. Esther was also pivotal to the development of replica plating. This technique enables scientists to make a perfect copy of the geometric pattern of all bacterial colonies growing on an agar plate. Her method revolutionised the ability to screen bacteria for a desired mutation and to track their mutations. Before her invention, scientists spent many painstaking hours transferring bacteria between petri dishes with blotting paper, toothpicks, or a wire brush. Coming from a family of textile workers, Esther grasped that velvet cloth with the right pile thickness offered a much simpler solution because its dense fibres could act as tiny needles to pick up and transfer the bacterial colonies in one go. By attaching the fabric to a wooden block, Esther showed that it could pick up just enough bacteria from a master agar plate to create identical colonies on other plates. Esther worked for decades at Stanford University, USA, becoming director of its Plasmid Reference Center in 1976. Her innovative work in microbial genetics proved an inspiration for many other researchers. Esther M Zimmer Lederberg (1922–2006) © 2021 Esther M Zimmer Lederberg Memorial Website/estherlederberg.com 2021 The pathologist and bacteriologist Mary Barber is another overlooked pioneer. Working at the Hammersmith Hospital in London, UK, Barber was one of the first to document the rise of antibiotic resistance and its role in cross infection in hospitals. An effective mobiliser, she showed that it was possible to reduce drug-resistant staphylococcal infections in hospitals by getting all the hospital staff to adhere to strict hygiene and sterilisation procedures and limiting antibiotics only to patients who would benefit. Predicting in 1950 that antibiotic-resistant staphylococcal infection would soon become an urgent problem, Barber's pioneering measures are an important lesson for today's control of hospital-acquired infections. The contributions of women during the COVID-19 pandemic and in the past are a powerful reminder of their pioneering roles behind advances in biomedical science. Running throughout the stories of the women featured here is their determination, collaborative approach, and ingenuity, which are vital ingredients to making progress in science and medicine. Productivity, innovation, decision making, and employee retention are enhanced when women are given an equal role in organisations and gender diversity is supported. Institutions that place emphasis on diversity and inclusion also tend to outperform those that do not. Sarah Gilbert © 2021 Steve Parsons/PA Wire/PA Images 2021 Katalin Karikó © 2021 ZUMA Press, Inc/Alamy Stock Photo 2021 Kizzmekia Corbett © 2021 Tim Nwachukwu/The New York Times via Getty Images 2021 Sharon Peacock I am the managing editor of the charitable educational platform WhatisBiotechnology.org, a Visiting Research Fellow at the Jeffrey Cheah Biomedical Centre, and the curator of an online exhibition about the history of AMR and scientists' struggles to overcome the problem. I am grateful to Kizzmekia Corbett, Sarah Gilbert, Katalin Karikó, and Sharon Peacock for taking the time to read and comment on an earlier version of this essay. ==== Refs Further reading Barber M Dutton AA Beard MA Reversal of antibiotic resistance in hospital staphylococcal infection BMJ 1 1960 11 17 13796582 Bevan V Coles C Why the UK lacks an adequate testing system The Guardian April 16, 2020 Bourbeau PP Burnham CAD What happened to research in clinical microbiology in the United States? J Clin Microbiol 49 suppl 9 2011 S85 S89 Corbett K The duty to mentor, be visible and represent Nat Med 26 2020 1670 Faulconbridge G UK's top COVID-19 virus hunter had a long and winding path to the top March 18, 2021 Reuters https://www.reuters.com/article/uk-health-coronavirus-britain-peacock-ne-idUSKBN2BA26G Garda D Saltzman J The story of mRNA: how a once-dismissed idea became a leading technology in the Covid vaccine race Nov 10, 2020 STAT https://www.statnews.com/2020/11/10/the-story-of-mrna-how-a-once-dismissed-idea-became-a-leading-technology-in-the-covid-vaccine-race/ Garnett C Corbett recounts quest for Covid vaccine. NIH Rec https://nihrecord.nih.gov/2020/12/11/corbett-recounts-quest-covid-vaccine Dec 11, 2020 Hesse W Walther and Angelina Hesse—early contributors to bacteriology ASM News 58 1992 425 428 Jefferson L Bloor K Maynard A Women in medicine: historical perspectives and recent trends Br Med Bull 114 2015 5 15 25755293 Kirchhelle C Dougan G Make it new: reformism and British public health Lancet Microbe 1 2020 e231 e232 32901232 Lane R Sarah Gilbert: carving a path towards a COVID-19 vaccine Lancet 395 2020 1247 Lovett S “You just get on with it”: the Oxford professor carrying the world's hopes of a coronavirus vaccine The Independent March 4, 2021 Ludwig M Big Pharma's blindspot: before COVID-19, vaccine and antiviral research went neglected Salon April 25, 2020 Nature COVID is amplifying the inadequacy of research-evaluation processes Nature 591 2021 7 33658700 Marks L The history of antimicrobial resistance and scientists' struggles to overcome the problem What Is Biotechnology? November, 2020 https://www.whatisbiotechnology.org/index.php/exhibitions/antimicrobial OSI The backstory: Vaccitech and its role in co-inventing the Oxford COVID-19 vaccine https://www.oxfordsciencesinnovation.com/news/the-backstory-vaccitech-and-its-role-in-co-inventing-the-oxford-covid-19-vaccine/ Nov 23, 2020 Piqueras M Esther Lederberg, pioneer of bacterial genetics. American Society for Microbiology Blog https://schaechter.asmblog.org/schaechter/2014/07/esther-lederberg-pioneer-of-bacterial-genetics.html July 28, 2018 Saul H International Women's Day: scientists leading us through Covid-19 on their work, careers, and women in science iNews March 8, 2021 Shannon G Jansen M Williams K Gender equality in science, medicine, and global health: where are we at and why does it matter? Lancet 393 2019 560 569 30739691 Further viewing Horizon Special: The Vaccine Produced and directed by Catherine Gale https://wingspanproductions.co.uk/what-we-do/read/69/Horizon-Special-The-Vaccine 2021

34303428

PMC9753770

NO-CC CODE

2022-12-16 23:26:18

no

Lancet. 2021 Jul 22 24-30 July; 398(10297):293-295

utf-8

Lancet

10.1016/S0140-6736(21)01647-0

oa_other

==== Front Lancet Lancet Lancet (London, England) 0140-6736 1474-547X Elsevier Ltd. S0140-6736(21)01197-1 10.1016/S0140-6736(21)01197-1 World Report Yemen's health system has “collapsed”, warns UN Devi Sharmila 29 5 2021 29 May-4 June 2021 29 5 2021 397 10289 20362036 © 2021 Elsevier Ltd. All rights reserved. 2021 Elsevier Ltd Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. ==== Body pmcThe ongoing conflict is exacerbating the effects of COVID-19 as aid agencies warn of a worsening humanitarian situation. Sharmila Devi reports. Yemen's health system has “collapsed” under the COVID-19 pandemic, with shortages of oxygen and personal protective equipment adding to humanitarian needs—including nutrition, water, and sanitation—for millions of people, the UN has warned. “[COVID-19] tests remain in short supply, aid agencies in Yemen are operating on the basis that community transmission is taking place across the country, and only half of the health facilities are fully functioning”, Jens Laerke, spokesperson for the UN Office for the Coordination of Humanitarian Affairs, said in a press briefing on May 22. “Yemen's health system needs significant assistance to counter the threat.” He warned that more than 30 key UN programmes risked closing in the coming weeks because of a lack of funding, including COVID-19 rapid response teams that were funded only for the next 6 weeks. “If we do not get the money coming in, the programmes that are keeping people alive and are very much essential to fight back against COVID-19 will have to close”, he said. “And then the world will have to witness what happens in a country without a functioning health system battling COVID-19.” Up to US$2 billion was needed for this year, and the UN and Saudi Arabia will co-host a virtual pledging event on June 2 to support fundraising. Yemen's conflict, which has killed more than 100 000 people and almost destroyed the health system since 2015, has already caused what the UN says is the world's worst humanitarian crisis. 24 million Yemenis, or 80% of the population, are reliant on aid. Diseases such as cholera, diphtheria, dengue, measles, and malaria, as well as a polio outbreak, have hit the country in recent years. 25% of the population, including 2·1 million children and 1·2 million pregnant and lactating women, have either moderate or severe malnutrition, says WHO. Nearly 1·5 million families depend on food assistance to survive, the majority of whom have no visible means of support. Thousands more people have cancer, diabetes, or other chronic conditions, for which treatment is limited. In 2014, rebel Houthis, known officially as the Ansar Allah movement, took control of Yemen's north and captured the capital Sanaa, forcing the UN-recognised government there to flee to Aden. Since 2015, a Saudi-led coalition of mostly Arab countries has been battling the rebels to reinstate control. As of May 23, 2021, there were 6653 confirmed COVID-19 cases and 1305 deaths reported to WHO, and 18 555 vaccine doses administered. However, extrapolation of the results of a WHO seroprevalence study in Aden indicated “that the infected number of cases is way more than those reported”, the WHO Yemen Country Office said in an email to The Lancet on May 20. “COVID-19 has placed additional pressure on existing facilities and resulted in underuse of health services, complicating the delivery of services to prevent other diseases. There will be growing food insecurity and limited hygiene and sanitation services, particularly for women and children”, WHO said in the email. The pandemic was further compounded by a continued lack of public awareness and adherence to protective measures, such as physical distancing and mask wearing, Marc Schakal, programme manager for Médecins Sans Frontières (MSF) in Yemen, told The Lancet. “Last year, during the first wave, there was a lot of panic and staff didn't come into work at public hospitals, forcing some to close. This hasn't happened during this second wave, but still staff are paid very irregularly, if at all, with a big impact on morale”, he said. “Also, some hospitals are turning suspected COVID-19 patients away, including pregnant women due for delivery, just because they have COVID-like symptoms.” More would be known about the trajectory of the pandemic in about 2 weeks, he said. There had been a decrease in admissions to MSF facilities during Ramadan, which ended in mid-May, but normal daily life and circulation of people has now resumed. Schakal also worried about a spillover effect from donor cuts. “We've already seen some [non-governmental organisations] decreasing their presence and more people coming into our centres from much further away, so our catchment area is increasing”, he said. Adding to the humanitarian burden are Saudi military measures preventing ships from delivering supplies. In the USA, in a letter dated May 20, a group of Democratic senators urged US President Joe Biden to take “immediate and decisive action” to end what they called Saudi Arabia's “blockade tactics” that have prevented food, medicine, and other crucial deliveries.

34062136

PMC9753771

NO-CC CODE

2022-12-16 23:26:18

no

Lancet. 2021 May 29 29 May-4 June; 397(10289):2036

utf-8

Lancet

10.1016/S0140-6736(21)01197-1

oa_other

==== Front Thromb Res Thromb Res Thrombosis Research 0049-3848 1879-2472 Elsevier Ltd. S0049-3848(21)00461-8 10.1016/j.thromres.2021.09.009 Full Length Article Changes in the epidemiology of patients hospitalized in France with deep venous thrombosis and pulmonary embolism during the COVID-19 pandemic Gabet Amélie a⁎ Grave Clémence a Tuppin Philippe b Emmerich Joseph c Olié Valérie a a Santé Publique France, Saint-Maurice 94, France b Caisse Nationale d'Assurance Maladie, Paris c Department of Vascular Medicine, Hôpital Saint-Joseph, Université de Paris, INSERM CRESS 1153, Paris, France ⁎ Corresponding author at: 14 rue du Val d'Osne, 94410 Saint Maurice, France. 21 9 2021 11 2021 21 9 2021 207 6774 20 7 2021 7 9 2021 13 9 2021 © 2021 Elsevier Ltd. All rights reserved. 2021 Elsevier Ltd Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. Introduction The onset of the COVID-19 pandemic and the first national lockdown implemented might have disrupted the epidemiology of deep venous thrombosis (DVT) and pulmonary embolism (PE). This study aimed to analyze time trends in patients hospitalized for DVT and PE in France and related in-patient and 90-day post-admission mortality rates. Materials and methods All patients hospitalized in France for DVT or PE between January and September (weeks 1–40) for each year from 2017 to 2020, were selected. Weekly incidence rate ratios (IRR) were computed to compare the rates of patients hospitalized in 2020 with those hospitalized in 2017–2019. Results Compared with the 2017–2019 study period, the rates of patients hospitalized with a primary diagnosis (PD) of DVT or PE in 2020 were significantly (50 and 40%, respectively) lower during weeks 12–13. The rate of patients hospitalized with an associated diagnosis (AD) of PE during weeks 12–19 of 2020 was twice as high as in the same period in 2017–2019. The prevalence of COVID-19 in patients hospitalized with a PD of DVT and PE, and in those hospitalized with an AD of DVT and PE reached respectively 4.0, 9.6, 17.2 and 44.6 during the country's first lockdown. Inpatients case-fatality rates in patients hospitalized with an AD of PE increased significantly during weeks 12–13. Conclusions Epidemiology of VT and PE was seriously impacted by the COVID-19 pandemic in 2020 in France, with a significant decrease in the rate of patients hospitalized for PE and a threefold increase in the related in-patient mortality rate. This highlight the need to inform the general population about the symptoms of PE and about the need to immediately seek medical care, particularly those infected with SARS-CoV-2. Keywords Deep venous thrombosis Pulmonary embolism COVID-19 Epidemiology Mortality ==== Body pmc1 Introduction The epidemiology of deep venous thrombosis (DVT) and pulmonary embolism (PE) may have been strongly impacted not only by its direct association with SARS-CoV-2 infection [1], [2], [3], [4], but also by the obstacles to primary and hospital care brought about by the COVID-19 pandemic [5], [6], [7], [8]. More specifically, the global context of the pandemic has led to a change in the medical management for both inpatient and outpatient DVT and PE cases, leading to longer delays before access to healthcare structures [9], [10], [11], [12]. The first lockdown in France between 17 March and 11 May 2020 disrupted DVT and PE healthcare management systems [13]. This lockdown was very restrictive, as the entire population had to stay at home and could only move within 1 km of their residence. Furthermore, as in other countries, a global decrease in the number of all-cause hospitalizations was observed, the main hypothesis for this being that people were afraid of getting infected with SARS-CoV-2, and accordingly, delayed seeking medical care [14]. A higher risk of DVT and PE was found in patients infected with SARS-CoV-2 in 2020 [1], [2], [3], [4]. Several mechanisms might explain this, including severe viral pulmonary infection, inflammatory syndrome added to the usual risk factors of DVT, and increased levels of D-dimers, suggesting substantial coagulation activation and a hypercoagulable state [15]. Although many international studies have analyzed DVT and PE incidence and prevalence in COVID-19 patients [1], [2], [4], to date none has compared the time trends in the rates of hospitalized DVT and PE patients during the COVID-19 pandemic - and specifically during a lockdown period - with the rates from previous years. This study aimed to analyze the impact of the COVID-19 pandemic on the epidemiology of hospitalizations for DVT and PE in France - with a special focus on the first lockdown period - in terms of the rates of patients hospitalized in 2020 versus 2017–19, their characteristics, care management, and related in-patient and 90-day (i.e., from hospital admission) case-fatality rates. A secondary objective aimed at describe the prevalence of COVID-19 among patients hospitalized with DVT and PE, respectively. 2 Methods 2.1 Data source For this retrospective population-based cross-sectional study, all patients hospitalized with DVT or PE in France between January and October (i.e., weeks 1 to 40) each year from 2017 to 2020, were selected using the French national hospital discharge databases (Programme de médicalisation des systèmes d'information – PMSI). The PMSI exhaustively record all hospital stays in both the private and public healthcare sectors in France, and is part of the larger French national healthcare database (Systéme National des données de Santé-SNDS) which collects data on reimbursements for healthcare expenditures (drug treatments, medical procedures, specific care for long-term diseases, etc.) for the whole French population (approximatively 67 million people) [16]. The healthcare system in France guarantees partial or full reimbursement for all medical care. The SNDS records disease codes for diagnosed cases using the international classification of diseases 10th revision (ICD-10). For the present study, DVT was defined as having any of the following codes: I80-I82 (except I80.0), O22.3, O87.1, whether reported as a primary (PD) or associated (AD) diagnosis. For PE, the codes were I26 or O88.2, again irrespective of PD or AD. For both illnesses, we analyzed patients with PD and AD separately. A PD of DVT or PE meant that the respective disease was the reason for the hospitalization or admission to a medical department. An AD meant that DVT or PE was a comorbidity associated with a PD of an illness considered more severe than DVT or PE - whether or not this illness was COVID-19 - and which required specific acute care. For all the patients in the study, we also searched the SNDS database for a diagnosis of COVID-19, either during the index hospitalization for DVT/PE, or during a previous hospitalization for a non-DVT/PE condition occurring after 1 January 2020. Hospital diagnoses of COVID-19 cases were based on positive biological tests (polymerization chain reaction, antigen or serology tests) or on CT-scans showing pulmonary lesions specific to SARS-CoV-2 virus infection. COVID-19 hospitalization incidence started to increase exponentially in February 2020 in France [17]. The first wave of the epidemic peaked during week 13. Thereafter, incidence decreased and remained low until September 2020 (week 27) when it started to increase again, to later become the second wave [18]. 2.2 Data collection The following patient characteristics were collected from the databases: age, sex, medical history of venous thromboembolism (i.e., DVT or PE), stroke, coronary heart disease, heart failure, arrhythmia, valvulophathy, and any circulatory disease recorded in the five years before the index hospitalization for DVT/PE. Data were also collected on healthcare system reimbursements for hypertension, diabetes and hypercholesterolemia medication payments, as well as for other medications (antiplatelets, oral anticoagulants) with at least three deliveries in the 12 months preceding the DVT/PE index hospitalization. Using these data, cardiovascular and non-cardiovascular comorbidities were grouped according to the Charlson index score [19]. Three groups were defined according to the number of comorbidities as follows: 0–1 comorbidity group, 2–3 comorbidities group, and 4 or more comorbidities group. The inpatient case-fatality rate was recorded, as was the 90-day post-admission case-fatality rate, for DVT/PE patients whose mortality data could be correctly matched with hospital data (i.e. 99% of patients), and who were members of France's general health insurance scheme. Seventy-seven percent of our population's study were covered by this scheme, which is consistent with its overall percentage of the French population [16]. The evaluation of the 90-day case-fatality rate was restricted to these patients, as this rate is not recorded exhaustively by the other insurance schemes which cover the remaining 23% of France's population. 2.3 Statistical analysis We described characteristics of patients hospitalized for DVT and PE for 2020 and the 2017–2019 period. Comparisons between both time periods for each of the two diseases were performed using the Chi2 test for categorical variables, and the Wilcoxon-Mann-Whitney test for quantitative variables. Means and standard deviations (SD) were calculated for each quantitative characteristic. Weekly incidence rate ratios (IRR) based on a Poisson distribution were computed to model the rates of patients hospitalized for DVT and PE in 2020 compared with 2017–2019. Weekly odds ratios (OR) were used to assess changes in patients' characteristics or case-fatality rates between 2017 and 2019 and 2020. Weekly OR for inpatient and 90-day post-admission case-fatality rates were only computed for patients hospitalized with a PD or AD of PE, as PE - unlike DVT - is directly involved in the process of death. All weekly indicators were based on the week of the index hospitalization for both diseases, and were adjusted for potential time trends between 2017 and 2020. Sensitivity analyses were conducted after excluding patients who were diagnosed with COVID-19 either during the index hospitalization for DVT/PE, or during a prior hospitalization after 1 January 2020 for a different condition (whether COVID-19 or not). Analyses were performed using the SAS Enterprise Guide 7.1 software tool. 2.4 Ethics approval As per French governmental regulations and the National Ethics Committee's rules of practice, no patient consent was required for the present study. The databases used in the study contained pseudonymized patient information. 2.5 Data availability Full access to the SNDS is granted to the National Agency for Public Health (Santé Publique France) by governmental decree (regulatory decision DE-2011-078). The data underlying this article cannot be shared publicly. 2.6 Permission information The authors do hereby declare that all illustrations and figures in the manuscript are entirely original and do not require reprint permission. 3 Results Between week 1 (January) and week 40 (October) of 2020, a total of 121,225 individuals were hospitalized for DVT or PE. Of these, 9574 and 35,446 had a PD of DVT and PE, respectively, while 52,446 and 23,759 had an AD of DVT and PE, respectively (Table 1 ). The mean total number of patients hospitalized for both diseases during the same period in 2017–2019 was 113,106, demonstrating a combined overall increase of 7.2% in 2020 (+8119 cases).Table 1 Number of patients hospitalized for diagnosed VT and PE during the study time-period of each year from 2017 to 2020, and total proportion of hospital-based COVID-19 diagnoses in 2020. Table 1 N % diagnosed with COVID-19 while in hospitala 2017 2018 2019 2020 During a prior hospitalization During the index VT/PE hospitalization only Overall Deep venous thrombosis Primary diagnosis  W1-W40 (January–October) 12,789 11,952 11,294 9574 0.9 0.6 1.5  W1-W11 (pre-lockdown) 3940 3614 3333 3167 0.1 0.2 0.3  W12-W19 (lockdown) 2553 2443 2421 1474 1.2 2.8 4.0  W20-W40 (post-lockdown) 6296 5895 5540 4933 1.2 0.3 1.5 Associated diagnosis, % of all VT diagnoses 78.6% 80.3% 82.2% 84.6%  W1-W40 (January–October) 46,963 48,975 52,311 52,446 1.0 3.6 4.6  W1-W11 (pre-lockdown) 14,879 14,949 15,818 16,694 0.1 0.7 0.8  W12-W19 (lockdown) 9216 9434 10,470 9501 1.9 15.3 17.2  W20-W40 (post-lockdown) 22,868 24,592 26,023 26,251 1.3 1.0 2.3 Pulmonary embolism Primary diagnosis  W1-W40 (January–October) 32,439 33,016 33,158 35,446 1.0 2.7 3.7  W1-W11 (pre-lockdown) 10,285 10,618 10,197 9810 0.0 0.4 0.4  W12-W19 (lockdown) 6059 6520 6691 6855 1.8 7.8 9.6  W20-W40 (post-lockdown) 16,095 15,878 16,270 18,781 1.2 1.9 3.1 Associated diagnosis, % of all PE diagnoses 16.7% 16.7% 16.4% 19.6%  W1-W40 (January–October) 18,541 18,895 18,985 23,759 1.7 14.7 16.4  W1-W11 (pre-lockdown) 6194 6356 6131 6279 0.2 2.4 2.6  W12-W19 (lockdown) 3461 3654 3629 6453 3.4 41.2 44.6  W20-W40 (post-lockdown) 8886 8885 9225 11,027 1.6 5.9 7.5 DVT: Deep venous thrombosis; PE: Pulmonary embolism; W: week. a Diagnosis based on hospital biological test (PCR, antigen or serology tests) or CT scan. 3.1 Time-trends and profile of patients hospitalized with DVT and PE as primary diagnosis The number of patients hospitalized with a PD of DVT during France's first lockdown (weeks 12–19) in 2020 was almost half that for the same period in 2017–2019 (N2020 = 1474 versus N2017 = 2553, N2018 = 2443, N2019 = 2421). By contrast, the numbers for PE remained relatively stable (N2020 = 6855 versus N2017 = 6059, N2018 = 6520, N2019 = 6691) (Table 1). With regard to weekly IRR (Fig. 1 , A and B), very contrasting time-trends were observed, with decreases of 50% and almost 40% in DVT and PE, respectively, during the first two weeks of the lockdown (weeks 12–13). Lower hospitalization rates for DVT persisted until two weeks after the lockdown ended (week 20–21, Fig. 1, A). From week 22 until week 40 (i.e., the end of the study period), no significant IRR was observed meaning that hospitalization rates for DVT were similar to those for the same period (i.e., week 22–40) in 2017–2019.Fig. 1 Weekly incidence rate ratios* (IRR) of patients hospitalized with deep venous thrombosis (DVT) and pulmonary embolism (PE) in 2020 compared with the corresponding weeks in 2017–2019, according to type of diagnosis and excluding COVID-19 diagnosis during hospitalization (whether prior or index DVT/PE hospitalization) Vertical axis: IRR; horizontal axis: week of hospitalization; grey zone: first national lockdown (weeks 12–19) *adjusted for time-trends between 2017 and 2019 and 2020. Fig. 1 The pattern was different however for patients hospitalized with a PD of PE (Fig. 1, B). More specifically, although rates fell by almost 40% in weeks 12–13, they quickly increased in the following weeks of the lockdown (weeks 13–16), this increase becoming significant in weeks 17–20, where hospitalization was 30% higher than for the same period in 2017–2019 (Fig. 1, B). Furthermore, unlike the result for DVT, between weeks 20–40 of 2020, rates were regularly higher than in 2017–2019 (Fig. 1, B). The percentage of patients hospitalized with a PD of DVT and PE in 2020 who were diagnosed in hospital with COVID-19 was 1.5 and 3.7%, respectively, reaching 4.0 and 9.6%, respectively, during the first lockdown (Table 1). For those diagnosed with COVID-19 during a prior hospitalization, the median delay before that hospitalization and the index hospitalization with a PD of DVT and PE was 46 (IQR: 14–130 days) and 20 (IQR: 8–65 days) days, respectively. Patients with a PD of DVT or PE who were diagnosed with COVID-19 were more frequently men and were younger (data not shown). In the sensitivity analysis, the exclusion of patients diagnosed with COVID-19 either during the index hospitalization for DVT/PE, or during prior hospitalization, did not modify these results (Fig. 1, C and D). Compared with 2017–2019, patients hospitalized in 2020 with a PD of DVT/PE were slightly younger (64.7 and 68.5 years old, respectively, versus 65.6 and 69.0 years old, p < 0.01), and were more likely to be in the 0–1 comorbidity group (in terms of the Charlson index score), to have a history of PE/DVT, and a history of medication for hypertension and hypercholesterolemia (Table 2 ).Table 2 Characteristics of patients hospitalized for VT and PE, and of the related hospital stay between January (Week 1) and October (Week 40) 2020 compared with the corresponding time period in 2017–2019, according to the type of diagnosis. Bold: p-value < 0.05. Table 2 Deep venous thrombosis as primary diagnosis Pulmonary embolism as primary diagnosis Deep venous thrombosis as associated diagnosis Pulmonary embolism as associated diagnosis Year 2017–2019 2020 p 2017–2019 2020 p 2017–2019 2020 p 2017–2019 2020 p Patients' characteristics Women, % 52.8 52.4 0.50 52.6 52.0 0.05 48.2 46.6 <0.01 50.6 47.3 <0.01 Age, mean (std) 65.6 (19.4) 64.7 (19.9) <0.01 69.0 (16.8) 68.5 (17.2) <0.01 67.0 (18.8) 66.9 (18.6) 0.18 70.3 (16.0) 69.4 (16.1) <0.01 0–65 years old, % 41.6 42.9 0.03 33.9 35.1 <0.01 38.0 37.9 0.79 31.1 33.6 <0.01 Charlson index score, % 0.02 <0.01 <0.01 <0.01  • 0–1 42.5 43.5 49.8 50.5 38.1 38.2 35.2 41.5  • 2–3 27.6 26.2 27.5 27.7 31.9 32.4 33.9 31.6  • 4+ 29.9 30.3 22.7 21.8 30.0 29.4 30.8 26.9 Hospitalization or official ‘long-term disease’ status in the five years preceding the index hospitalizationa  • Pulmonary embolism 3.6 2.7 <0.01 10.0 9.3 <0.01 2.6 1.7 <0.01 13.4 9.6 <0.01  • Venous thromboembolism 16.9 17.8 0.04 12.2 10.9 <0.01 11.4 9.8 <0.01 16.0 11.4 <0.01  • Coagulation abnormalities 9.8 10.1 0.42 6.2 6.3 0.52 7.8 7.6 0.07 6.4 5.3 <0.01  • Stroke 5.8 5.9 0.52 5.6 5.7 0.36 8.1 8.4 0.03 9.7 9.4 0.36  • Ischemic heart disease 15.5 14.8 0.08 12.9 12.2 <0.01 16.8 17.2 0.06 15.4 14.1 <0.01  • Heart failure 13.7 13.2 0.21 14.0 13.4 0.01 15.4 15.6 0.13 16.6 14.2 <0.01 Treatment delivery in the year preceding the index hospitalizationa  • Antihypertensives 53.8 52.6 0.03 53.7 51.1 <0.01 57.3 57.3 0.77 58.1 54.9 <0.01  • Antidiabetics 15.7 16.2 0.27 12.0 11.9 0.69 17.6 18.7 <0.01 15.6 16.0 0.20  • Lipid-lowering medications 26.7 25.2 <0.01 27.0 24.4 <0.01 27.7 27.9 0.20 28.4 26.6 <0.01  • Oral anticoagulants 11.0 10.5 0.15 7.3 5.9 <0.01 14.9 14.7 0.16 14.8 11.1 <0.01  • Heparin 5.6 4.7 <0.01 2.6 2.1 <0.01 6.2 4.8 <0.01 5.2 3.5 <0.01  • Antiplatelets 24.5 23.2 <0.01 22.3 20.9 <0.01 25.1 25.1 0.96 24.2 23.0 <0.01 Disease management and hospital stay Length of stay, mean (std) 7.1 (9.0) 7.2 (9.9) <0.01 9.3 (9.4) 9.0 (9.0) <0.01 16.1 (16.8) 16.5 (17.3) <0.01 17.7 (18.2) 18.0 (17.9) <0.01 Admitted to intensive care unit, %b 4.4 4.8 0.09 26.7 26.1 0.02 16.9 19.8 <0.01 23.4 28.0 <0.01 In-patient mortality, % 1.7 1.8 0.53 4.8 4.9 0.51 7.3 7.8 <0.01 15.3 16.1 <0.01 90-day post-admission mortality, %b 7.6 8.3 0.06 7.2 8.0 <0.01 12.0 11.9 0.82 15.4 15.0 0.27 std: standard deviation; p: 2020 vs. 2017–2019. a Available for patients without linkage error (i.e., 99% of patients). b Available for patients affiliated to France's general healthcare insurance scheme (i.e., 77% of included patients). 3.2 Time-trends and profile of patients hospitalized with DVT and PE as an associated diagnosis Although the number of patients hospitalized with an AD of DVT in weeks 12–19 (i.e., the first lockdown) in 2020 remained stable with respect to 2017/2019, the number of patients hospitalized with an AD of PE doubled (weeks 12–19: N2020 = 6453) (N2017 = 3461; N2018 = 3654; N2019 = 3629) (Table 1). Weekly IRR differed between patients hospitalized with an AD of DVT and PE (Fig. 1, E and F), and were also different from those found for patients hospitalized with a PD of DVT and PE (described in the previous section) (Fig. 1, A and B). In terms of hospitalization rates for an AD of PE, a large and significant increase was observed throughout all of the first lockdown (weeks 12–19) in 2020, with respect to the same period in 2017–2019. Specifically, the weekly hospitalization rate for weeks 13–17 was twice as high (Fig. 1, F). The rates of hospitalizations for an AD of DVT or PE were higher for patients under 75 years old (versus those over 75 years old) (See supplemental Figure e1). On the contrary, a slight but significant decrease in hospitalization rates with an AD of DVT was observed during the lockdown (Fig. 1, E). This decrease was only significant in people over 75 years old (See supplemental Fig. e2). The percentage of patients with an index hospitalization with an AD of DVT and PE in 2020 who were diagnosed in hospital with COVID-19 was 4.6 and 16.4%, respectively, reaching 17.2 and 44.6% during the first lockdown (weeks 12–19). Specifically, for those diagnosed with COVID-19 during a prior hospitalization, the median delay before that hospitalization and the index hospitalization for an AD of DVT/PE was shorter than that for a PD of DVT/PE (see preceding section), at 26 (IQR: 7–86 days) and 10 (IQR: 4–31 days) days, respectively. In the sensitivity analysis, the exclusion of patients diagnosed with COVID-19 either during the index hospitalization for DVT/PE or during a prior hospitalization on or after 1 January 2020, did not modify these results (Fig. 1, G and H). Compared with 2017–2019, the population of patients hospitalized with an AD of PE in 2020 were more likely to be men (52.7% vs 40.4%, p < 0.01), to be in the 0-1 comorbidity group (41.1% vs 34.7%, p < 0.01), and to be admitted to an intensive care unit (ICU) (28.0% vs 23.4%, p < 0.01) at some point during the hospitalization. They also had a higher inpatient case-fatality rate (16.1% vs 15.3%, p < 0.01). Similarly, the patient profile of persons hospitalized with an AD of DVT changed between 2017 and 2019 and 2020 (See supplemental Fig. e2), with a large increase in the proportion of men and in persons under 65 years old, and a substantial decrease in the proportion of patients with at least two comorbidities (i.e., the 2–3 and 4 or more comorbidity groups). These trends were greater for PE than DVT (See supplemental Fig. e2). 3.3 In-hospital and 90-day post-admission mortality in patients hospitalized with a primary or associated diagnosis of PE Compared with the corresponding time period in 2017–2019 (weeks 1–40), the OR for in-patient case-fatality rates did not change significantly for hospitalizations with a PD of PE in 2020 (Fig. 2 , A and B). On the contrary, inpatients hospitalized in 2020 with an AD of PE were, respectively, three times more likely to die than their counterparts hospitalized in 2017–2019 during the first week of lockdown (week 13, Fig. 2, C). After excluding cases diagnosed with COVID-19 in hospital, significantly higher case-fatality rates were still observed for week 12 of 2020 with respect to the same week in 2017–2019 patients hospitalized with an AD of PE (Fig. 2, D).Fig. 2 Weekly adjusted* odds ratios (OR) of in-patient hospital case-fatality rates among patients hospitalized with a pulmonary embolism (PE) in 2020 compared with the corresponding weeks in 2017–2019, according to the type of diagnosis and COVID-19 diagnosis during hospitalization Vertical axis: IRR; horizontal axis: week of hospitalization; grey zone: first national lockdown (weeks 12–19) *adjusted for time-trends between 2017 and 2019 and 2020, age and sex. Fig. 2 The 90-day post-admission case-fatality rates significantly increased between the two study periods, but in an irregular fashion; specifically, increases were seen for weeks 11 and 17 for persons with a PD of PE (See supplemental Fig. e3, A and B) and for weeks 5, 18 and 28 for those with an AD of PE (See supplemental Fig. e3, C). These observations remained after excluding patients diagnosed with COVID-19 in hospital (See supplemental Fig. e3, B and D). 4 Discussion Our study showed that significant changes occurred in the epidemiology of hospitalized DVT and PE in France in 2020 compared with 2017–2019. This was particularly true during the country's first lockdown (weeks 12–19), with a significant decrease being observed in hospitalizations of people with a PD of DVT and PE by 50 and 40%, respectively, in the first two weeks (weeks 12–13), and a significant increase in hospitalizations for those with an AD of PE, reaching up to 200% during the first five weeks of the lockdown (weeks 12–17). During the first lockdown, 10 and 45% of persons hospitalized with a PD and an AD of PE, respectively, were diagnosed with COVID-19 during hospitalization (whether index DVT/PE hospitalization or prior hospitalization). The inpatient case mortality rate was three times higher in persons hospitalized with an AD of PE in the first week of the lockdown (week 12) in 2020 than in their hospitalized counterparts during the corresponding week in 2017–2019. The relative decrease (i.e., compared with the same period in 2017/2019) in the rates of patients hospitalized with a PD of DVT and PE during the first two weeks of the lockdown (weeks 12–13) in 2020, was consistent with the decline observed in France and elsewhere in hospitalizations for other acute cardiovascular diseases - such as stroke [20], [21] and acute myocardial infarction [22] - and with the global decrease in the use of healthcare services [23]. The main hypothesis explaining these time-trends is that people were afraid of getting COVID-19 in hospital and therefore decided to delay contacting healthcare services. Nevertheless, the effect of this on PE was short-lived, as the numbers of hospitalizations started to increase again as early as the second week of the lockdown (week 13), and returned to normal (i.e., compared with the mean level for 2017/2019) during week 15. The decrease in both DVT and PE hospitalizations during the first lockdown may also be partly linked to cancellations by health authorities of non-urgent surgery (mainly orthopedic and digestive surgery), and to the decrease in trauma surgery in that period [24], [25]. Between the end of the first lockdown (week 19) and the end of the study period (week 40), the weekly hospitalization rates for people with a PD of PE were often higher in 2020 than the mean for 2017–2019. This may be due to PE occurring as a consequence of an increase in COVID-19 cases. This point underlines the importance of extending thrombo-prophylactic measures to outpatient COVID-19 cases at high risk of PE [26]. Aktaa et al. showed a substantial increase in mortality from PE without COVID-19 diagnosis during the first wave of the pandemic in the UK. The authors linked this to the lack of testing outside hospitals during this period [27]. In terms of hospitalizations of people with a PD of DVT, the lower rates observed at the beginning of the lockdown persisted during and after the entire lockdown period until week 21. The difference in the patterns observed for DVT and PE might be explained by the fact that a majority of patients diagnosed with DVT were provided care outside the hospital setting when the peak of the first wave of the pandemic in France occurred in March 2020. The increase in the numbers of patients hospitalized with an AD of PE in France during the whole first lockdown period was directly linked to the explosion in the numbers of SARS-CoV-2 infections in the general population, which resulted in mass hospitalizations for COVID-19 and related PE [1], [2], [4]. This finding is consistent with the increase in the rates of patients hospitalized with venous thromboembolism events in the abovementioned UK study [27]. Many studies have described the association between COVID-19 and PE. In ours, 45% of patients hospitalized with an AD of PE during France's first lockdown were diagnosed with COVID-19. Therefore, at that time, the epidemiology of PE and COVID-19 converged, which explained the substantial increase in the proportion of men, younger patients, and ICU admissions (data not shown for this last one) we found in patients hospitalized with an AD of PE. We cannot exclude the possibly - as hypothesized in the UK study above [27] - that the incidence of venous thromboembolism increased as a consequence of people staying at home too long without doing any physical activity. The substantial increase in inpatient case-fatality rates for persons hospitalized with an AD of PE in the first two weeks of France's first lockdown (weeks 12–13) in 2020, might be related to the difficulty to manage patients who have both a severe form of COVID-19 and PE, and to the greater severity of COVID-19-related PE cases [28]. The rapid implementation of anticoagulation therapy and, when necessary, mechanical thromboprophylaxis (as per several national and international recommendations [26], [29], [30], [31]) in hospitalized COVID-19 cases with a severe form of the disease - especially those admitted to ICU - together with the greater awareness by healthcare professionals of the high risk of thromboembolic events in severe COVID-19 patients, might have reduced PE incidence in severe COVID-19 cases and therefore limited the inpatient case-fatality rate. The increase (with respect to 2017–2019) in adjusted DVT and PE inpatient case-fatality rates which we saw in 2020, may also be related to longer delays before accessing hospital services during the first two weeks (weeks 12–13) of the first lockdown, whether because people waited longer before seeking medical attention, or because hospitals were too overwhelmed. This increase is consistent with the global excess in mortality observed in France in the initial weeks of the first lockdown [32]. A return to normality (with respect to 2017–2019) in adjusted DVT and PE inpatient case-fatality rates was observed from week 14 until the end of July 2020 in patients who were diagnosed with COVID-19. This may be related to better management of COVID-19 patients with thrombosis (thanks to widespread diffusion of effective recommendations [26], [29]), to the decline in the COVID-19 epidemic at that point in time, to the end of the first lockdown (and therefore possibly faster access to hospital), and to less fear about getting the disease in hospital (and therefore seeking care earlier) [18]. However, the reasons for the subsequent increase in inpatient case-fatality rates after July 2020 in this same population remain unclear. 4.1 Public health and clinical implications Our results highlight that in the context of the ongoing COVID-19 pandemic and venous thromboembolism, the related lockdown measures to contain the pandemic, the general public's fear of getting COVID-19 if they go to hospital (resulting in a delay before seeking medical attention), and the structural delays in being able to access healthcare structures, all need to be tackled by increasing the general population's awareness of the severity of venous thromboembolism and its complications, especially if medical care is not sought immediately after symptoms appear. The association between DVT after prolonged sitting was first recognized during the blitzkrieg in World War II, when fatal PE were observed in Londoners obliged to stay for long periods of times in air shelters [33]. The restrictive effects of COVID-19-related lockdowns on people's movements, within and outside of their homes, highlights a parallel situation and similar risk. Furthermore, the increased risk of venous thromboembolism in patients with COVID-19 which led to increased hospitalizations for DVT and PE in France in 2020, highlights the need to: (a) adjust the quantity and type of resources allocated to the management of venous thromboembolism in hospitals, (b) ensure adapted anticoagulation therapy in patients with COVID-19 at risk of DVT [26], [29], and (c) advocate for closer follow-up of these patients. Further research is also necessary to evaluate whether the spread of new SARS-CoV-2 variants in the ongoing pandemic has differentially impacted the incidence of DVT in France and elsewhere. 4.2 Strengths and limitations To the best of our knowledge, this is the first nationwide study to exhaustively investigate the epidemiology of hospitalized DVT and PE cases, as well as associated case-fatality rates in France over a relatively long and recent time period, as cases until October 2020 were included. The main limitation of the study was the unavailability of data on outpatient DVT and PE cases, and associated case-fatality rates, in a context where related inpatient mortality in France increased by 9% with respect to 2017–2019 [32], [34]. A second limitation is that the time between DVT/PE symptom onset and hospitalization was not available in the database used, which limits the interpretation of our results. Finally, another limitation of our study was the general use of administrative medical databases which were not conceived for epidemiological but economic purposes. 5 Conclusion The present study showed that the epidemiology of DVT and PE was seriously impacted by the COVID-19 pandemic in 2020. First, measures taken to limit population movements, especially lockdowns, were associated with a 40% decrease in the rate of patients hospitalized for PE with respect to 2017–2019, and a threefold increase in the related in-patient mortality rate, although the effect did not last for a long time. Second, COVID-19 itself directly led to an increase in the numbers of patients hospitalized with DVT and PE in France. This finding highlights the importance of promoting anticoagulation therapy in COVID-19 patients at risk of DVT and advocating for close follow-up of COVID-19 patients at risk of PE. More generally, efforts to increase the general population's awareness of DVT and PE need to be intensified, and the quantity and type of inpatient resources devoted to the management of these diseases need to be increased. Funding sources None. Declaration of competing interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Appendix A Supplementary data Supplementary figures Image 1 Acknowledgment None. Appendix A Supplementary data to this article can be found online at https://doi.org/10.1016/j.thromres.2021.09.009. ==== Refs References 1 Suh Y.J. Hong H. Ohana M. Bompard F. Revel M.P. Valle C. Pulmonary embolism and deep vein thrombosis in COVID-19: a systematic review and meta-analysis Radiology 298 2 2021 E70 e80 10.1148/radiol.2020203557 33320063 2 Di Minno A. Ambrosino P. Calcaterra I. Di Minno M.N.D. COVID-19 and venous thromboembolism: a meta-analysis of literature studies Semin. Thromb. Hemost. 46 7 2020 763 771 10.1055/s-0040-1715456 32882719 3 Helms J. Tacquard C. Severac F. Leonard-Lorant I. Ohana M. Delabranche X. High risk of thrombosis in patients with severe SARS-CoV-2 infection: a multicenter prospective cohort study Intensive Care Med. 46 6 2020 1089 1098 10.1007/s00134-020-06062-x 32367170 4 Jiménez D. García-Sanchez A. Rali P. Muriel A. Bikdeli B. Ruiz-Artacho P. Incidence of VTE and bleeding among hospitalized patients with coronavirus disease 2019: a systematic review and meta-analysis Chest 159 3 2021 1182 1196 10.1016/j.chest.2020.11.005 33217420 5 Georger F. Dos Santos E. Gazagne L. Berdagué P. Saib A. Nahon S. COV IMPACT: stress exposure analysis among hospital staff in 2 hospitals in France during the COVID-19 pandemic Ann. Cardiol. Angeiol. 69 5 2020 227 232 10.1016/j.ancard.2020.09.005 6 Danguy des Déserts M. Mathais Q. Luft A. Escarment J. Pasquier P. Conception and deployment of a 30-bed field military intensive care hospital in Eastern France during the 2020 COVID-19 pandemic Anaesth. Crit. Care Pain Med. 39 3 2020 361 362 10.1016/j.accpm.2020.04.008 32360981 7 Kuteifan K. Pasquier P. Meyer C. Escarment J. Theissen O. The outbreak of COVID-19 in Mulhouse : hospital crisis management and deployment of military hospital during the outbreak of COVID-19 in Mulhouse, France Ann. Intensive Care 10 1 2020 59 10.1186/s13613-020-00677-5 32430597 8 Lefrant J.Y. Fischer M.O. Potier H. Degryse C. Jaber S. Muller L. A national healthcare response to intensive care bed requirements during the COVID-19 outbreak in France Anaesth. Crit. Care Pain Med. 39 6 2020 709 715 10.1016/j.accpm.2020.09.007 33031979 9 Rosovsky R.P. Grodzin C. Channick R. Davis G.A. Giri J.S. Horowitz J. Diagnosis and treatment of pulmonary embolism during the coronavirus disease 2019 pandemic: a position paper from the national PERT consortium Chest 158 6 2020 2590 2601 10.1016/j.chest.2020.08.2064 32861692 10 Davin-Casalena B. Jardin M. Guerrera H. Mabille J. Tréhard H. Lapalus D. The impact of the COVID-19 pandemic on first-line primary care in southeastern France: feedback on the implementation of a real-time monitoring system based on regional health insurance data Rev. Epidemiol. Sante Publique 69 3 2021 105 115 10.1016/j.respe.2021.04.135 33992499 11 Fournier J.P. Amélineau J.B. Hild S. Nguyen-Soenen J. Daviot A. Simonneau B. Patient-safety incidents during COVID-19 health crisis in France: an exploratory sequential multi-method study in primary care Eur. J. Gen. Pract. 27 1 2021 142 151 10.1080/13814788.2021.1945029 34212814 12 Lesieur O. Quenot J.P. Cohen-Solal Z. David R. De Saint Blanquat L. Elbaz M. Admission criteria and management of critical care patients in a pandemic context: position of the Ethics Commission of the French Intensive Care Society, update of April 2021 Ann. Intensive Care 11 1 2021 66 10.1186/s13613-021-00855-z 33904016 13 Cauchemez S. Kiem C.T. Paireau J. Rolland P. Fontanet A. Lockdown impact on COVID-19 epidemics in regions across metropolitan France Lancet (London, England). 396 10257 2020 1068 1069 10.1016/s0140-6736(20)32034-1 33007219 14 Morello F. Bima P. Ferreri E. Chiarlo M. Balzaretti P. Tirabassi G. After the first wave and beyond lockdown: long-lasting changes in emergency department visit number, characteristics, diagnoses, and hospital admissions Intern. Emerg. Med. 1–8 2021 10.1007/s11739-021-02667-2 15 Page E.M. Ariëns R.A.S. Mechanisms of thrombosis and cardiovascular complications in COVID-19 Thromb. Res. 200 1–8 2021 10.1016/j.thromres.2021.01.005 16 Tuppin P. Rudant J. Constantinou P. Gastaldi-Menager C. Rachas A. de Roquefeuil L. Value of a national administrative database to guide public decisions: from the systeme national d'information interregimes de l'Assurance maladie (SNIIRAM) to the systeme national des donnees de sante (SNDS) in France Rev. Epidemiol. Sante Publique 65 Suppl 4 2017 S149 s67 10.1016/j.respe.2017.05.004 28756037 17 Bernard Stoecklin S. Rolland P. Silue Y. Mailles A. Campese C. Simondon A. First cases of coronavirus disease 2019 (COVID-19) in France: surveillance, investigations and control measures, January 2020 Euro. Surveill. 25 6 2020 2000094 10.2807/1560-7917.ES.2020.25.6.2000094 18 Santé Publique France Point Epidémiologique du 8 Octobre 2020 https://www.santepubliquefrance.fr/maladies-et-traumatismes/maladies-et-infections-respiratoires/infection-a-coronavirus/documents/bulletin-national/covid-19-point-epidemiologique-du-8-octobre-2020 2021 19 Quan H. Li B. Couris C.M. Fushimi K. Graham P. Hider P. Updating and validating the charlson comorbidity index and score for risk adjustment in hospital discharge abstracts using data from 6 countries Am. J. Epidemiol. 173 6 2011 676 682 10.1093/aje/kwq433 21330339 20 Gabet A. Grave C. Tuppin P. Chatignoux E. Béjot Y. Olié V. Impact of the COVID-19 pandemic and a national lockdown on hospitalizations for stroke and related 30-day mortality in France: a nationwide observational study Eur. J. Neurol. 2021 10.1111/ene.14831 21 Nogueira R.G. Abdalkader M. Qureshi M.M. Frankel M.R. Mansour O.Y. Yamagami H. Global impact of COVID-19 on stroke care Int. J. Stroke 16 5 2021 573 584 10.1177/1747493021991652 33459583 22 Mesnier J. Cottin Y. Coste P. Ferrari E. Schiele F. Lemesle G. Hospital admissions for acute myocardial infarction before and after lockdown according to regional prevalence of COVID-19 and patient profile in France: a registry study Lancet Public Health 5 10 2020 e536 e542 10.1016/s2468-2667(20)30188-2 32950075 23 Moynihan R. Sanders S. Michaleff Z.A. Scott A.M. Clark J. To E.J. Impact of COVID-19 pandemic on utilisation of healthcare services: a systematic review BMJ Open 11 3 2021 e045343 10.1136/bmjopen-2020-045343 24 Barro K. Malone A. Mokede A. Chevance C. Management of the COVID-19 epidemic by public health establishments-Analysis by the Fédération Hospitalière de France J. Visc. Surg. 157 3s1 2020 S19 s23 10.1016/j.jviscsurg.2020.04.011 32417194 25 Druel T. Andeol Q. Rongieras F. Bertani A. Bordes M. Alvernhe A. Evaluation of containment measures’ effect on orthopaedic trauma surgery during the COVID-19 pandemic: a retrospective comparison between 2019 and 2020 Int. Orthop. 44 11 2020 2229 2234 10.1007/s00264-020-04712-9 32696333 26 Khider L. Soudet S. Laneelle D. Boge G. Bura-Rivière A. Constans J. Proposal of the French Society of Vascular Medicine for the prevention, diagnosis and treatment of venous thromboembolic disease in outpatients with COVID-19 J. Med. Vasc. 45 4 2020 210 213 10.1016/j.jdmv.2020.04.008 32571561 27 Aktaa S. Wu J. Nadarajah R. Rashid M. de Belder M. Deanfield J. Incidence and mortality due to thromboembolic events during the COVID-19 pandemic: multi-sourced population-based health records cohort study Thromb. Res. 202 17–23 2021 10.1016/j.thromres.2021.03.006 28 Planquette B. Le Berre A. Khider L. Yannoutsos A. Gendron N. de Torcy M. Prevalence and characteristics of pulmonary embolism in 1042 COVID-19 patients with respiratory symptoms: a nested case-control study Thromb. Res. 197 94–9 2021 10.1016/j.thromres.2020.11.001 29 Huisman M.V. Klok F.A. Kaptein F.H.J. Stals M.A.M. Prophylaxis and treatment of COVID-19 related venous thromboembolism Postgrad. Med. 2021 1 9 10.1080/00325481.2021.1891788 30 Susen S. Tacquard C.A. Godon A. Mansour A. Garrigue D. Nguyen P. Prevention of thrombotic risk in hospitalized patients with COVID-19 and hemostasis monitoring Crit. Care 24 1 2020 364 10.1186/s13054-020-03000-7 32560658 31 Fontana P. Casini A. Robert-Ebadi H. Glauser F. Righini M. Blondon M. Venous thromboembolism in COVID-19: systematic review of reported risks and current guidelines Swiss Med. Wkly. 150 2020 w20301 10.4414/smw.2020.20301 32 Fouillet A. Pontais I. Caserio-Schönemann C. Excess all-cause mortality during the first wave of the COVID-19 epidemic in France, March to May 2020 Eurosurveillance 25 34 2020 2001485 10.2807/1560-7917.ES.2020.25.34.2001485 33 Simpson K. Shelter deaths from pulmonary embolism Lancet 236 6120 1940 744 10.1016/S0140-6736(00)92078-6 34 Institut national de la statistique et des études économiques 2020 : une hausse des décès inédite depuis 70 ans. Insee Première Mars 2021 1847 https://www.insee.fr/fr/statistiques/5347349

34562677

PMC9753774

NO-CC CODE

2022-12-16 23:26:18

no

Thromb Res. 2021 Nov 21; 207:67-74

utf-8

Thromb Res

10.1016/j.thromres.2021.09.009

oa_other

==== Front Tech Coloproctol Tech Coloproctol Techniques in Coloproctology 1123-6337 1128-045X Springer International Publishing Cham 36520243 2741 10.1007/s10151-022-02741-7 Original Article Pelvic floor physical therapy in the treatment of chronic anal fissure (PAF trial): quality of life outcome http://orcid.org/0000-0002-7861-1915 van Reijn-Baggen D. A. davr@me.com 14 Elzevier H. W. 2 Braak J. P. B. M. 3 Putter H. 5 Pelger R. C. M. 4 Han-Geurts I. J. M. 1 1 Department of Surgery, Proctos Clinics, Bilthoven, The Netherlands 2 grid.10419.3d 0000000089452978 Department of Urology and Medical Decision Making, Leiden University Medical Center, Leiden, The Netherlands 3 grid.10419.3d 0000000089452978 Department of Surgery, Leiden University Medical Center, Leiden, The Netherlands 4 grid.10419.3d 0000000089452978 Department of Urology, Leiden University Medical Center, Leiden, The Netherlands 5 grid.10419.3d 0000000089452978 Department of Biomedical Data Science, Leiden University Medical Center, Leiden, The Netherlands 15 12 2022 19 23 6 2022 5 12 2022 © Springer Nature Switzerland AG 2022, Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. Background Chronic anal fissure is one of the most common anorectal diseases and is associated with reduced quality of life. The aim of this study was to investigate the effects of pelvic floor physical therapy on quality of life in patients with chronic anal fissure using the Short-Form 36 Health Survey (RAND-36). Methods Adult patients, with chronic anal fissure and concomitant pelvic floor dysfunction, such as dyssynergia and increased pelvic floor muscle tone, were recruited at the Proctos Clinic in the Netherlands, between December 2018 and July 2021 and randomly assigned to an intervention group, receiving 8 weeks of pelvic floor physical therapy or assigned to a control group receiving postponed pelvic floor physical therapy (PAF trial). Quality of life and pain ratings were outcomes of the study and were measured at 8- and 20-week follow-up. Results One hundred patients (50 women and 50 men, median age 44.6 years [range 19–68 years]), completed the RAND-36 questionnaire and visual analog (VAS) pain scale score at admission. A significant improvement was found at 20-week follow-up in all domains of the RAND-36; physical functioning, pain, health change (p < 0.001); physical role, vitality, general health, social functioning, emotional role, mental health (p < 0.05). VAS pain was significantly reduced at 8 weeks (mean estimated difference 1.98; 95% CI 1.55–2.42, p < 0.001) and remained significant at 20-week follow-up (p < 0.001). The difference between the groups as regards change in the mean pain intensity scores at 8 weeks was 2.48 (95% CI − 3.20 to − 1.75; p < 0.001). Compared to the reference values of the general Dutch population, the patients in our study with a chronic anal fissure and pelvic floor dysfunction reported an impaired quality of life in 8 of 9 domains of the RAND-36. After treatment, significant lower scores were found in 2 out of 9 domains. Conclusions The results of this study provide evidence that treatment by pelvic floor physical therapy improves quality of life and reduces pain, making it an important tool in management of chronic anal fissure and concomitant pelvic floor dysfunction. Keywords Chronic anal fissure Quality of life QoL RAND-36 SF-36 Pelvic floor physical therapy ==== Body pmcIntroduction Chronic anal fissure (CAF) is a common proctological problem associated with reduced quality of life [1]. CAF is defined as a longitudinal ulcer in the squamous epithelium [2] and gives rise to distressing symptoms of pain and bleeding during and after defecation. The incidence of CAF is nearly 0.11% (1.1 cases per 1000 persons) and varies considerably according to age and sex [3]. Persistence of symptoms for long periods may lead to functional and psychosocial impairment [4], and seeking medical care is often delayed due to embarrassment [5]. Furthermore, in patients with CAF, there is a high degree of depression, anxiety and stress [1]. Health-related quality of life (QoL) can be influenced by physical, psychological and social factors, an individual’s life experience and general well-being [1, 6]. The purpose of health-related QoL evaluations is to move beyond clinical symptoms by examining how patients perceive and experience the impact on well-being and daily life [6, 7]. The most common generic instrument to measure QoL is the validated Medical Outcomes 36-Item Short-Form Health Survey (SF-36) used for decision-making for health care policies and clinical interventions [8]. Although there is a need to integrate aspects of functional and psychosocial impairment into medical care [9], only a few studies report on QoL in patients with CAF. Recently, the Pelvic floor Anal Fissure study (PAF trial) was completed, which is a randomized controlled trial demonstrating the beneficial effects of pelvic floor physical therapy (PFPT) on an improvement of pelvic floor muscle tone and function, VAS pain scores, fissure healing and complaint reduction [10]. The aim of PFPT is to increase awareness and proprioception, to improve muscle relaxation, elasticity and function of the pelvic floor muscles, to restore abdominopelvic coordination, rectal sensitivity and reduce pain [11, 12]. In the PAF trial, we also hypothesized that treatment of PFPT will improve QoL. Here, we present the results of QoL measured with the Short-Form RAND-36 (RAND-36) [13] and visual analog scale (VAS) pain ratings in patients with CAF and pelvic floor dysfunction, who were included in the PAF trial. Furthermore, to better elucidate the results, the study compares baseline and post-treatment values with reference values of the RAND-36 of the general Dutch population [13]. Materials and methods Study design Quality of life was assessed with the RAND-36 in the PAF trial [14]. The PAF trial is a single-center, parallel, randomized controlled trial. The design involved allocation of all appropriate consecutive patients older than 18 years with CAF and pelvic floor dysfunction. Eligible patients were randomly assigned, after providing written informed consent to an intervention group receiving 8 weeks of PFPT or assigned to postponed PFPT (1:1 allocation). Participants Men and women aged 18 years or older presenting CAF and pelvic floor dysfunction were recruited by the surgeon at the Proctos Clinic in the Netherlands. CAF was defined as a longitudinal ulcer in the squamous epithelium with one or more signs of chronicity including hypertrophied anal papilla, sentinel tag and exposed internal sphincter muscle. Patients had fissure complaints of more than 6 weeks, and all patients failed in conservative treatment with fibers and/or laxatives and had applied the ointment (diltiazem or isosorbide di-nitrate) internally for at least 6 weeks. Pelvic floor dysfunction was defined by the presence of dyssynergia and/or increased pelvic floor muscle tone. All patients had sufficient understanding of the Dutch language (reading and writing) and were able to complete the online questionnaires. Patients who were not able to undergo a digital rectal examination, patients with an abscess or fistula, Crohn’s disease or ulcerative colitis, anorectal malignancy, previous rectal or anal surgery, previous rectal radiation and pregnant patients were excluded from the trial. Physical examination and questionnaires The diagnosis of CAF was based on the patient’s medical history and a thorough local inspection of the anus. Resting anal sphincter pressure was measured by a careful digital rectal examination and scored as normal, weak (decreased), or increased [15]. Pelvic floor dysfunction was defined by the presence of dyssynergia and/or increased pelvic floor muscle tone. Pelvic floor muscle tone was measured with a digital rectal examination [16] and surface electromyography (μV) [16] with an intra-anal probe (Maple, ®Novuqare Pelvic Health B.V. CE 0344, Rosmalen, the Netherlands), which is validated for its purpose [17]. Pelvic floor dyssynergia was detected by digital rectal examination [18] and balloon expulsion test [19]. If necessary, proctoscopy was performed to exclude other pathology. To access the impact of global QoL, the validated Dutch version of Short-Form RAND-36, Health Status Inventory, version 2 [13] was used. The RAND-36 consists of 36 items and 9 subscales: physical functioning, bodily pain, role limitation due to physical health problems, vitality, general health perception, social functioning, role limitation due to emotional problems, mental health, and health change perception. The RAND-36 consists of the same sets of items as the SF-36) [20], although the scoring procedure differs between the RAND-36 and SF-36 for the domains of general health and bodily pain. The score for each scale is obtained by the sum of the scores for each item linearly transformed into a range from 0 to 100. A higher score indicates more favorable QoL. To quantify the average intensity of pain during defecation, a VAS from 0 (no pain) to 10 (most intense pain) was used [21]. Patients were requested to fill in the RAND-36 and VAS score at baseline, and at 8- and 20-week follow-up. Interventions At baseline, patients in both groups received information about the pelvic floor and related symptoms, explanation about relevant anatomy and defecation (patho)physiology, behavioral modifications and lifestyle advice. All patients continued their conservative measures including the use of ointment (diltiazem or isosorbide di-nitrate). PFPT consisted of 5 face-to-face appointments of 45 min in a period of 8 consecutive weeks, using a treatment protocol. Details of this treatment protocol were described earlier [14]. Patients who were assigned to postponed PFPT did not receive additional treatment besides their conservative measures and the use of ointment until first follow-up at 8 weeks after inclusion. Patients from the postponed PFPT group followed the same treatment protocol after first follow-up. Data collection of the RAND-36 was facilitated by a secure online system called Castor EDC [22]. Patients received the questionnaire by e-mail through the Castor system at 3 time points: at baseline, at 8- and 20-week follow-up. Outcome measures The primary outcome measure of this study was QoL in patients with CAF and pelvic floor dysfunction before and after PFPT and compared to reference values of the general Dutch population. The other outcome measure was the average pain intensity during defecation on a VAS-scale. The sample size of the PAF-study was based on the primary endpoint, the tone at rest during electromyographic registration of the pelvic floor and consisted of 140 patients [14]. The data from the questionnaires that were at least 75% completed at baseline and follow-up were used for analysis. Statistical analysis Data were analyzed using Statistical Packages for Social Sciences (SPSS, Chicago, II, USA, version 28.0). Descriptive methods were used to assess quality of data, homogeneity of treatment groups and endpoints. Normality of the data was analyzed with histograms. Data are presented using mean (SD), median (range) for the numeric and non-normal variables and frequency (percentages) for categorical variables. A paired t test or Wilcoxon signed-rank test was used to compare continuous variables within groups. An independent T test or Mann–Whitney U test for quantitative data was performed to analyze statistical differences between groups. For each of the dimensions of the RAND-36, items scores were coded, summed and transformed on to a scale of 0–100. Statistical analyses consisted of estimating means and standard deviations for each of the RAND-36 scale scores. Comparison between groups for continuous variables was made by repeated measure analysis of variance using a mixed model after transformation of the data to enhance normality, with treatment, time (categorical) and their interaction as fixed effects and with random patient effects. To acquire an indication of the QoL of life of patients with CAF as compared to the reference group of the Dutch population, we calculated for each dimension the significance from the norm score [13] with the one-sample t test. In case of missing data, we excluded that specific case from further analyses when less than 75% of the questionnaire was filled out. All p values were two-tailed, and statistical significance was taken as a p value of less than 0.05. Results Between December 2018 and July 2021, 140 patients were randomized to PFPT or postponed PFPT. After randomization, 3 patients withdrew. The RAND-36 was adequately completed by 100 patients at baseline, of whom 50 women and 50 men with a median age of 44.6 years (range 19–68 years). The results from the questionnaires at baseline of 37 patients were excluded because less than 75% of the form had been completed. The participants’ demographic and clinical characteristics of the total group of patients from the PAF-study, those who completed the baseline questionnaire adequately and individual treatment groups, are presented in Table 1. There were no significant differences in terms of demographic or clinical parameters between the groups at baseline (Table 1).Table 1 Baseline demographics Variable Total group PAF-study (n = 140) Total with adequate baseline data RAND-36 (n = 100) PFPT RAND-36 (n = 52) Postponed PFPT RAND-36 (n = 48) Age, years, median (range) 44.5 (19–79) 44.6 (19–68) 44.4 (23–66) 44.8 (19–68) Sex: women/men (%) 51.4/48.6 50/50 53.8/46.2 45.8/54.2 Duration of complaints (%)  0–2 months 12.1 15.0 13.5 16.7  2–6 months 22.9 26.0 25.0 27.1  6–12 months 14.3 12.0 13.5 10.4  12–36 months 22.1 26.0 25.0 27.1  > 3 years 28.6 21.0 23.1 18.8  VAS pain score (mean, SD) 5.3 ± 1.6 5.5 ± 1.7 5.6 ± 1.6 5.4 ± 1.8 VAS visual analog scale, PFPT pelvic floor physical therapy, RAND-36 short-form 36 health survey The non-response rate at 20-week follow-up was 31%. The results of the mean RAND-36 sub-scores from the different domains and the mean VAS pain scores, per time point from the total group and individual treatment groups, are presented in Table 2.Table 2 Study measures at baseline, 8-week and 20 -week follow-up. Comparison within and between treatment groups and repeated measurements for all patients with adequate baseline data Quality of life scale RAND-36 Total group PFPT Postponed PFPT MD between groups Baseline mean (SD) (n = 100) 20 weeks mean (SD) (n = 69) p valueΣ Baseline mean (SD) (n = 52) 8 weeks mean (SD) (n = 45) p value∞ 20 weeks mean (SD) (n = 37) p valueΣ Baseline mean (SD) (n = 48) 8 weeks mean (SD) (n = 40) p value∞ 20 weeks mean (SD) (n = 32) p valueΣ p value∞ Physical functioning 82.9 (20.9) 90.3 (14.1) < 0.001¢ 84.9 (18.7) 89.2 (17.1) p = 0.124¢ 91.8 (14.0) p = 0.052¢ 80.7 (23.0) 87.1 (16.9) p = 0.012¢ 88.6 (14.3) p = 0.008¢ p = 0.174π Bodily pain 55.0 (26.5) 76.8 (20.5) < 0.001τ 54.8 (25.2) 69.7 (24.7) < 0.001τ 78.1 (19.6) < 0.001τ 55.3 (28.2) 74.4 (21.9) < 0.001¢ 75.4 (21.6) < 0.001τ p = 0.339ψ Physical role 56.0 (44.1) 76.1 (38.9) p = 0.004¢ 57.7 (43.3) 78.9 (35.7) p = 0.004¢ 79.7 (38.6) p = 0.022¢ 54.2 (45.4) 73.1 (36.4) p = 0.012¢ 71.9 (39.5) p = 0.049¢ p = 0.176π Vitality 55.5 (18.7) 60.9 (16.1) p = 0.013τ 55.5 (20.4) 58.7 (19.5) p = 0.296τ 61.1 (16.8) p = 0.054τ 55.5 (16.9) 57.6 (15.6) p = 0.283τ 60.6 (15.6) p = 0.130τ p = 0.753ψ General health 66.9 (18.8) 69.9 (17.6) p = 0.036τ 71.2 (17.7) 73.2 (19.3) p = 0.394¢ 73.1 (18.4) p = 0.484τ 62.3 (19.0) 64.9 (17.8) p = 0.222τ 66.3 (16.2) p = 0.029τ p = 0.469ψ Social functioning 72.4 (24.2) 81.9 (22.2) p = 0.011¢ 70.2 (23.6) 81.4 (22.2) p = 0.010¢ 84.5 (21.3) p = 0.031¢ 74.7 (24.8) 80.0 (22.3) p = 0.065¢ 78.9 (23.2) p = 0.157¢ p = 0.455ψ Emotional role 70.7 (41.4) 83.6 (31.1) p = 0.043¢ 76.3 (36.4) 82.9 (33.0) p = 0.085¢ 84.7 (28.9) p = 0.747¢ 64.6 (45.8) 79.2 (32.6) p = 0.072¢ 82.3 (33.9) p = 0.047¢ p = 0.192π Mental health 66.2 (14.5) 70.9 (13.0) p = 0.003τ 65.9 (14.6) 71.1 (17.7) p = 0.050τ 71.5 (11.5) p = 0.042τ 66.5 (14.7) 70.5 (14.0) p = 0.035τ 70.4 (14.8) p = 0.038τ p = 0.701ψ Health change 46.5 (24.6) 62.3 (23.3) < 0.001¢ 49.5 (26.4) 61.7 (28.5) p = 0.004τ 64.2 (25.4) p = 0.003τ 43.2 (22.3) 55.6 (24.3) p = 0.004τ 60.2 (20.9) < 0.001τ p = 0.726ψ VAS pain score 5.5 (1.7) 1.5 (1.6) < 0.001¢ 5.6 (1.6) 2.4 (2.0)  < 0.001¢ 1.4 (1.6) < 0.001¢ 5.4 (1.8) 4.7 (1.9)  < 0.001¢ 1.6 (1.6) < 0.001¢  < 0.001ψ RAND-36 short-form 36 health survey, PFPT pelvic floor physical therapy, MD mean difference τPaired t test ¢Wilcoxon signed-rank test πMann–Whitney U test ψUnpaired T test ∞p value 8 weeks vs baseline Σp value 20 weeks vs baseline QoL pre- and post-treatment For the group who adequately completed the questionnaire, the mean scores significantly improved in all domains of the RAND-36 from baseline to 20-week follow-up; physical functioning, bodily pain, health change (p < 0.001); physical role, vitality, general health, social functioning, emotional role and mental health (p < 0.05) (Table 2, Fig. 1). Fig. 1 Median Short-Form 36 Health Survey (RAND-36) scores of the total group before and after treatment QoL pre- and post-treatment for individual treatment groups At 8-week follow-up, the PFPT group had significantly improved as regards bodily pain (p < 0.001), physical role, social functioning, mental health, and health change (p < 0.05) and the effect remained significant at 20-week follow-up. No significant improvement was found in vitality, general health and emotional role at 8- and 20-week follow-up (Table 2). The postponed PFPT group significantly improved in the domains, bodily pain (p < 0.001), physical functioning, physical role, mental health, and health change (p < 0.05) at 8-week follow-up and remained significant at 20-week follow-up. At 20 weeks, the postponed group significantly improved in general health and emotional role (p < 0.05) post treatment. No significant improvements were found in the postponed group in the domain vitality and social functioning at this time point (Table 2). According to the mean estimated difference between groups at 8-week follow-up, no significant differences were found in the different domains of the RAND-36 (Table 2). Repeated measurement analysis showed more improvement in all domains in time from baseline to follow-up at 20 weeks in the PFPT group compared to postponed PFPT group although these differences were not significant (Fig. 2).Fig. 2 Repeated measurement analysis. PFPT pelvic floor physical therapy, CI confidence interval Pain For the group as a whole, the VAS was significantly reduced from baseline to follow-up at 8 weeks (mean estimated difference 1.98; 95% CI 1.55–2.42, p < 0.001) and remained significant at 20-week follow-up (p < 0.001) (Table 2). The VAS pain score was significantly reduced in both the PFPT and the postponed PFPT group at 8 weeks from baseline (p < 0.001). At 20-week follow-up, the VAS pain score in the PFPT group and postponed PFPT group further decreased and remained significant compared to baseline (p < 0.001). The difference between the groups as regards change in the mean pain intensity scores at 8 weeks from baseline was 2.48 (95% CI − 3.20 to − 1.75; p < 0.001) favoring the PFPT group. At 20 weeks, no significant mean difference in VAS scores was found between groups (p = 0.269). QoL in the total group compared to the Dutch population Compared to the reference group of the general Dutch population based on a mean age of 44 years [13], patients with CAF scored significantly lower on the subscales bodily pain, physical role, vitality, social functioning, mental health (p < 0.001) and general health, emotional role, and health change (p < 0.05). No significant difference was found in the domain physical functioning (p = 0.633) (Fig. 3).Fig. 3 Mean Short-Form 36 Health Survey (RAND-36) scores of the total group and the reference group from the Dutch population Results showed that patients had higher post-treatment scores at 20-week follow-up compared to the Dutch reference values on physical functioning and health change (p < 0.001), but the scores in vitality and mental health were still significantly lower (p < 0.001). No significant difference was found between the whole groups compared to the normal Dutch population on the other domains at this time point. Discussion Health-related QoL measured by the RAND-36 significantly improved in all dimensions in all patients at 20-week follow-up and confirm the efficacy of PPFT on quality of life in patients with CAF from the PAF trial. The literature on the RAND-36 shows that very small differences in the range of 3–5 points on the survey could be interpreted as clinically important [23, 24]. In all domains of the RAND-36, the minimal clinical importance was higher than 3 points, which could be interpreted as indicating that the treatment was meaningful to the patient. Furthermore, compared to the reference values of the general Dutch population, patients with CAF and pelvic floor dysfunction reported an impaired QoL in 8 of 9 domains of the RAND-36. After treatment, significantly lower scores were found in 2 out of 9 domains. The positive effect of PFPT on QoL in patients with other anorectal complaints [25, 26] is already known but has never been investigated in patients with CAF. In our study, the PFPT group significantly improved in 5 of 9 domains of the RAND-36 at 8-week follow-up. Interestingly, the postponed PFPT group also improved in 5 of 9 domains. An important aspect of treatment is re-education and improving understanding of defecation disorders [27]. It is likely that the information all patients receive about their complaints, instruction about toilet behavior and lifestyle advice also are reflected in an improvement in QoL in the postponed PFPT group, explaining our results. Neither group improved in the domains, general health, vitality and emotional role at 8- and 20-week follow-up. One of the reasons could be that RAND-36 is not sensitive enough to pick up changes in these domains in a relatively short period of time (i.e., 20 weeks). More studies with a long-term follow-up are needed to confirm this. In the domain bodily pain, all patients significantly improved post treatment compared to baseline. The same results were found for VAS pain scores. Reduction of pain is likely to have a positive reflection on QoL. Results from a study by Griffin et al. [4] in patients with CAF who were treated with topical ointment, confirm this assumption. Higher VAS pain scores were associated with worse outcome in all aspects of health-related QoL, with pain influencing many psychosocial and functional activities. A study by Tsunoda et al. [28] examining the treatment of CAF with diltiazem found that pain had a negative impact on the domains bodily pain and social functioning at baseline. Patients with healed fissures after treatment, reported an improvement in bodily pain, vitality, general health, and mental health. The PAF-study [10] found that the fissure was healed in 60% of all patients at 20-week follow-up. Significant lower scores were found in patients with non-healed fissures in the domains, bodily pain, social functioning, and emotional role at that time point. In a study by Bagul et al. [29] of patients with CAF who received Botulinum toxin injections, pain scores improved in 74% of the patients. QoL improved in patients in the domains of physical functioning, bodily pain, social functioning and mental health. The study demonstrated that pain was a significant factor influencing the outcome of QoL scores. Another study investigating Qol after lateral internal sphincterotomy in 58 patients [30] found improvement in pain symptoms although not all domains of health-related QoL were similarly positively affected. Smaller gains were reported among younger participants, women, participants with no comorbidities and those participants who waited the longest for their surgery. Patients with CAF in our study scored lower overall than the reference group of the Dutch population. One of the reasons could be the chronicity of the problem. In our population, 65% of the patients had complaints for more than 6 months, which would have a negative influence on the patient, family members and other relations [31]. Other factors influencing the outcome of treatment should be investigated in further studies with a long-term follow-up. The conclusions of this study are strengthened by the response rate of 71% at baseline, the high sample size and prospective design of the study. We enrolled patients of all ages and both sexes from different parts of the Netherlands. Thus, the results may be generalizable to the CAF population at large. This study has some limitations. Currently, there is no disease-specific tool for assessing QoL in patients with CAF and therefore a generic instrument was used. The RAND-36 was chosen because it is one of the most used questionnaires measuring QoL, and it is translated in Dutch [13]. Its reliability has been proven in a post-rehabilitation Dutch population [32] but may not be specific enough to fully analyze the QoL in patients with CAF. The non-response rate was 31% at 20-week follow-up. This may have caused bias if non-or partial respondents differ from respondents as concerns QoL or its determinants or confounders [33]. Reasons for non-completion at 20-week follow-up were surgery including Botulinum toxin, fissurectomy, fistulotomy, sclerodermy and other surgery (breastcancer). Other reasons were COVID-19, pregnancy, loss of follow-up for logistical reasons (distance, insurance, other) and personal. We did not find significant baseline differences between those followed up and those lost to follow-up. Although the results show a significant improvement in a short period of time (e.g., 20 weeks), it is unknown what the long-term outcome of PFPT on QoL will be. In the PAF trial, patients also visited the clinic at 1-year follow-up. At the time of submitting this manuscript, the results of the 1-year follow-up were not completed. Hence, they could not be incorporated. Conclusions The results of this study provide evidence that PFPT is effective in the improvement of QoL and positively influences pain in patients with CAF and pelvic floor dysfunction Patients with CAF and concomitant pelvic floor dysfunction reported an impaired QoL compared to the reference values of the general population in the Netherlands. Funding No funding of the study. Declarations Conflict of interest The authors declare that they have neither competing interests nor conflict of interest. Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References 1. Arısoy Ö Şengül N Çakir A Stress and psychopathology and its impact on quality of life in chronic anal fissure (CAF) patients Int J Colorectal Dis 2016 32 6 921 924 10.1007/s00384-016-2732-1 28039531 2. Nelson RL (2014) Anal fissure (chronic). BMJ Clin Evid 2014:0407. https://www.ncbi.nlm.nih.gov/pubmed/25391392. Accesssed 12 Nov 2014 3. Mapel DW Schum M Von Worley A The epidemiology and treatment of anal fissures in a population-based cohort BMC Gastroenterol 2014 14 129 10.1186/1471-230X-14-129 25027411 4. Griffin N Acheson AG Tung P Sheard C Glazebrook C Scholefield JH Quality of life in patients with chronic anal fissure Colorectal Dis 2004 6 1 39 44 10.1111/j.1463-1318.2004.00576.x 14692952 5. Gilani A Tierney G Chronic anal fissure in adults BMJ 2022 376 1756–1833 (Electronic) e066834 10.1136/bmj-2021-066834 35022226 6. Glise H Wiklund I Health-related quality of life and gastrointestinal disease J Gastroenterol Hepatol 2002 17 Suppl no. 0815-9319 (Print) S72 84 10.1046/j.1440-1746.17.s1.6.x 12000595 7. Fitzpatrick R Fletcher A Gore S Jones D Spiegelhalter D Cox D Quality of life measures in health care. I: Applications and issues in assessment, (in eng) BMJ 1992 305 6861 1074 1077 10.1136/bmj.305.6861.1074 1467690 8. Nilsson E Kristenson M Psychological factors related to physical, social, and mental dimensions of the SF-36: a population-based study of middle-aged women and men Patient Relat Outcome Meas 2010 1 1179–271X (Print) 153 162 10.2147/PROM.S13209 22915961 9. Mentes BB Tezcaner T Yilmaz U Leventoglu S Oguz M Results of lateral internal sphincterotomy for chronic anal fissure with particular reference to quality of life Dis Colon Rectum 2006 49 7 1045 1051 10.1007/s10350-006-0527-0 16598404 10. van Reijn-Baggen DA Elzevier HW Putter H Pelger RCM Han-Geurts IJM Pelvic floor physical therapy in patients with chronic anal fissure: a randomized controlled trial Tech Coloproctol 2022 26 7 571 582 10.1007/s10151-022-02618-9 35511322 11. van Reijn-Baggen DA Han-Geurts IJM Voorham-van der Zalm PJ Pelger RCM Hagenaars-van Miert C Laan ETM Pelvic floor physical therapy for pelvic floor hypertonicity: a systematic review of treatment efficacy Sex Med Rev 2022 10 2 209 230 10.1016/j.sxmr.2021.03.002 34127429 12. Bocchini R Pelvic floor rehabilitation for defecation disorders Tech Coloproctol 2019 23 2 101 115 10.1007/s10151-018-1921-z 30631977 13. VanderZee KI Sanderman R Heyink JW de Haes H Psychometric qualities of the RAND 36-Item Health Survey 1.0: a multidimensional measure of general health status Int J Behav Med 1996 3 2 104 122 10.1207/s15327558ijbm0302_2 16250758 14. van Reijn-Baggen DA Elzevier HW Pelger RCM Han-Geurts IJM Pelvic floor physical therapy in the treatment of chronic anal fissure (PAF-study): Study protocol for a randomized controlled trial Contemp Clin Trials Commun 2021 24 2451–8654 (Electronic) 100874 10.1016/j.conctc.2021.100874 34841124 15. Rao SSC Rectal exam: yes, it can and should be done in a busy practice! Am J Gastroenterol 2018 113 5 635 638 10.1038/s41395-018-0006-y 29453382 16. Frawley H An International Continence Society (ICS) report on the terminology for pelvic floor muscle assessment Neurourol Urodyn 2021 40 5 1217 1260 10.1002/nau.24658 33844342 17. Voorham-van der Zalm PJ Reliability and differentiation of pelvic floor muscle electromyography measurements in healthy volunteers using a new device: the Multiple Array Probe Leiden (MAPLe) Neurourol Urodyn 2013 32 4 341 348 10.1002/nau.22311 22972554 18. Tantiphlachiva K Rao P Attaluri A Rao SS Digital rectal examination is a useful tool for identifying patients with dyssynergia, (in eng) Clin Gastroenterol Hepatol 2010 8 11 955 960 10.1016/j.cgh.2010.06.031 20656061 19. Chiarioni G Kim SM Vantini I Whitehead WE Validation of the balloon evacuation test: reproducibility and agreement with findings from anorectal manometry and electromyography Clin Gastroenterol Hepatol 2014 12 12 2049 2054 10.1016/j.cgh.2014.03.013 24674941 20. The MOS 36-item short-form health survey (SF-36). I. Conceptual framework and item selection Med Care. 1992 Jun;30(6):473-83. 21. Dworkin RH Interpreting the clinical importance of treatment outcomes in chronic pain clinical trials: IMMPACT recommendations J Pain 2008 9 2 105 121 10.1016/j.jpain.2007.09.005 18055266 22. Castor Electronic Data Capture (EDC):https://www.castoredc.com. 23. Samsa G Edelman D Rothman ML Williams GR Lipscomb J Matchar D Determining clinically important differences in health status measures: a general approach with illustration to the health utilities index mark II Pharmacoeconomics 1999 15 2 141 155 10.2165/00019053-199915020-00003 10351188 24. Hays RD Morales LS The RAND-36 measure of health-related quality of life Ann Med 2001 33 5 350 357 10.3109/07853890109002089 11491194 25. Bartlett L Sloots K Nowak M Ho YH Supplementary home biofeedback improves quality of life in younger patients with fecal incontinence J Clin Gastroenterol 2015 49 5 419 428 10.1097/MCG.0000000000000143 25856243 26. Cadeddu F Salis F De Luca E Ciangola I Milito G Efficacy of biofeedback plus transanal stimulation in the management of pelvic floor dyssynergia: a randomized trial Tech Coloproctol 2015 19 6 333 338 10.1007/s10151-015-1292-7 25744688 27. Beaty JS Shashidharan M Anal Fissure, (in eng) Clin Colon Rectal Surg 2016 29 1 30 37 10.1055/s-0035-1570390 26929749 28. Tsunoda A Kashiwagura Y Hirose K Sasaki T Kano N Quality of life in patients with chronic anal fissure after topical treatment with diltiazem, (in eng) World J Gastrointest Surg 2012 4 11 251 255 10.4240/wjgs.v4.i11.251 23494072 29. Bagul A Carter P Quality of life in patients receiving botulinum a toxin treatment for symptomatic anal fissure World J Surg Surgical Res 2019 2 1130 30. Sutherland JA-O Health and quality of life among a cohort of patients having lateral internal sphincterotomy for anal fissures, (in eng) Colorectal Dis 2020 22 1463–1318 (Electronic) 1658 1666 10.1111/codi.15191 32533887 31. Navarro-Sánchez A Sexuality, quality of life, anxiety, depression, and anger in patients with anal fissure a case-control study LID 2021 10.3390/jcm10194401 32. Krops LA Wolthuizen L Dijkstra PU Jaarsma EA Geertzen JHB Dekker R Reliability of translation of the RAND 36-item health survey in a post-rehabilitation population, (in eng) Int J Rehabil Res 2018 41 2 128 137 10.1097/MRR.0000000000000265 29140827 33. Coste J Quinquis E Fau-Audureau L Audureau J Fau-Pouchot E Pouchot J Non response, incomplete and inconsistent responses to self-administered health-related quality of life measures in the general population: patterns, determinants and impact on the validity of estimates - a population-based study in France using the MOS SF-36 Health Qual Life Outcomes 2013 11 1477–7525 44 10.1186/1477-7525-11-44 23497315

36520243

PMC9753858

NO-CC CODE

2022-12-16 23:26:35

no

Tech Coloproctol. 2022 Dec 15;:1-9

utf-8

Tech Coloproctol

10.1007/s10151-022-02741-7

oa_other

==== Front Forensic Sci Med Pathol Forensic Sci Med Pathol Forensic Science, Medicine, and Pathology 1547-769X 1556-2891 Springer US New York 36520378 565 10.1007/s12024-022-00565-3 Original Article The effect of the COVID-19 pandemic on forensic cases admitted to an emergency department http://orcid.org/0000-0002-3790-9774 Sarı Doğan Fatma fatmasdogan@gmail.com http://orcid.org/0000-0002-4760-0076 Öztürk Tuba Cimilli Emergency Medicine Clinic, Fatih Sultan Mehmet Education and Research Hospital, Istanbul, Turkey 15 12 2022 16 21 11 2022 © The Author(s), under exclusive licence to Springer Science+Business Media, LLC, part of Springer Nature 2022, Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. Introduction: The COVID-19 disease has given rise to various negative effects on human life in terms of health and economic and social well-being. We believe that these negative effects may have led to increased forensic incidents such as violence and suicide. Therefore, in this study, we sought to examine the effects of COVID-19 in forensic cases admitted to an emergency department. Methods: This is a retrospective observational study, performed at the emergency department of Fatih Sultan Mehmet Education and Research Hospital. Forensic cases admitted between March and June 2020 (pandemic period) and forensic cases admitted between March and June 2019 (pre-pandemic period) were compared in the study. Results: A total of 4296 patients were included in the study, of which 3011 were admitted during the pre-pandemic period and 1285 during the time of the COVID-19 pandemic. While the percentages of suicide attempts (3.6%), motorcycle traffic accidents (7.4%), and violent incidents (29.4%) were higher during the pandemic period, the percentages of in-vehicle traffic accidents (5.4%) and pedestrian traffic accidents (2.2%) were lower (respectively, p = 0.035, p = 0.005, p < 0.001, p = 0.015, p = 0.008). At the time of the pandemic, the percentages of incidents of violence against women (44.2%) and traffic accidents with a motorcycle involving men (9.3%) were higher than during the time before the pandemic (p < 0.001 and p < 0.001, respectively). Conclusions: The effects of the pandemic on our lifestyle are indisputable. This study reveals that the pandemic also affected patients who were admitted to the emergency department for forensic reasons. In addition, the increase in the percentages of suicide and violent events indicates that pandemics probably increase feelings of fear, loss, and hopelessness, and special precautions should be taken to maintain order in the society. Keywords COVID-19 pandemic Forensic cases Emergencies Suicides Traffic accidents Incidence of violence ==== Body pmcIntroduction Incidents such as traffic accidents, poisonings, suicide attempts, assaults, and gunshot wounds that lead to the mental and physical deterioration and even death of individuals are considered judicial events. Diagnosis and medical intervention in forensic events are mostly done by emergency services [1–3]. The epidemic caused by the coronavirus COVID-19 has caused many changes in our lives. Along with the health problems caused by the disease, loss of relatives, restrictions in daily life, and financial difficulties, the risks of loneliness, hopelessness, insomnia, anxiety, anger, suicide, and violence have increased all over the world [4–6]. As a result of the measures taken, a decrease in traffic accidents can be predicted, due to the decrease in pedestrians and vehicles on the streets [7–9]. This study aims to compare the forensic cases admitted to our emergency department before and during the COVID-19 pandemic. Thus, we aim to reveal the effect of the COVID-19 pandemic on forensic cases. We believe that these data can shed light on preparing social support programs for the public in similar situations that may occur in the future. Materials and methods This research study was planned as a retrospective observational study and performed in the Emergency Department of Fatih Sultan Mehmet Education and Researh Hospital. Our hospital is a third-level training and research hospital in Istanbul, with 290,000 patients admitted to our emergency department annually. We define the pandemic period as 11.03.2020, when the first COVID-19 case was reported in our country, to 01.06.2020, when the normalization process started. We use the same date range 1 year previously to define the pre-pandemic period (11.03.2019 to 01.06.2019). Every patient admitted to the emergency department was questioned at our institution as a possible forensic case. Forensic cases were first evaluated in the emergency department by emergency medical specialists. This initial evaluation was based on the patient’s statement, a physical examination, and tests that can be performed under emergency conditions. Cases brought in by the police were also included. We selected the patients who were given a forensic case code from the hospital records. Regarding the inclusion and exclusion criteria, all cases given a forensic case code in the emergency department during the study period were included in the study. Although it was very unlikely, it was determined that those whose records had any missing data would be excluded from the study. Patients’ data were obtained from the hospital information system. Patients were grouped into pre-pandemic period cases and pandemic period cases. The age, gender, and reason for admission (in-vehicle traffic accident, pedestrian traffic accident, suicide attempt, assault, knife wound, firearm injury, work accident, forensic examination before or after police pursuit, fall from height, poisoning, drowning, burns, and electrical accident) were statistically compared between the groups. R Version 2.15.3 (R Core Team 2013) was used for the statistical analysis. The study data were used to report the means, standard deviations, frequencies, and percentages. The conformity of the quantitative data to the normal distribution was tested using the Shapiro–Wilk test and graphical examinations. An independent group t-test was used for comparisons between the two groups of normally distributed quantitative variables. Pearson’s chi-square test, Fisher’s exact test, and the Fisher-Freeman-Halton exact test were used to compare the qualitative data. Statistical significance was accepted as p < 0.05. Approval for the study was obtained from the Ministry of Health COVID-19 Scientific Research Platform, Number x-2020-06-04T11_03_30.xml, as well as from the Clinical Research Ethics Committee of Fatih Sultan Mehmet Education and Research Hospital, Number 2020/83. Results During the period determined for the study, 4296 patients were registered as forensic cases; all of these patients were included in the study, as there was no missing data. Of the total of 4296 patients included in the study, 3011 cases (70.08%) had been admitted during the pre-pandemic period, and 1285 (29.91%) during the pandemic period. Only 22.4% (n: 675) of the patients who were admitted during the pre-pandemic period and 26.1% (n: 335) of the patients who were admitted during the pandemic period were women. There was a significant difference between the pre-pandemic period and the pandemic period cases in terms of gender (p = 0.010). The rate of women who were admitted as forensic cases during the pandemic period was higher than before the pandemic. The mean age of pre-pandemic period admissions was 31.52 ± 12.66, while the mean age of pandemic period admissions was 32.13 ± 12.6 years. There was no statistically significant difference between the two groups in terms of age (p > 0.05). A total of 425 of the patients participating in the study were younger than 18 years of age, 122 during the COVID period, and 303 during the pre-COVID period. The cause-related distribution of the patients presenting during the pre-COVID period was examined: 40 patients (13.2%) were admitted due to traffic accidents, 155 patients (51.2%) for forensic examination, 82 patients (27.1%) due to assault, seven patients (2.3%) for suicide, 19 patients (6.3%) due to a work accident, four patients (1.3%) for falling from a height, and two (0.7%) patients for burns. The distribution of patients under the age of 18 at the time of COVID was as follows: seven patients (5.7%) were admitted due to a traffic accident, 80 patients (65.6%) for forensic examination, 28 patients (23%) due to assault, four patients (3.3%) for suicide, and three patients (2.5%) due to a work accident. During the pandemic, while the percentages of suicide attempts, motorcycle traffic accidents (TA), and assault incidents were higher than in the time before the pandemic, the percentages of in-vehicle TAs and pedestrian TAs were lower (respectively, p = 0.035, p = 0.005, p = 0.001, p = 0.015, p = 0.008). The percentage of women suffering domestic violence at the time of the pandemic was higher compared to the time before the pandemic (p < 0.001). For males, while the percentage of motorcycle TA events was higher at the time of the pandemic compared to the time before the pandemic, the percentages of forensic examinations and falling from height were lower (respectively, p < 0.001, p = 0.003, p < 0.001). The descriptive data are summarized in Table 1. The genders of the forensic cases, the pre-pandemic and pandemic distributions, and the p values are summarized in Table 2. Table 1 Descriptive values Total Pre-pandemic time cases Pandemic time cases p (n = 4296) (n = 3011) (n = 1285) Age, mean ± sd 31.70 ± 12.64 31.52 ± 12.66 32.13 ± 12.61 a0.148 Gender, n(%) b0.010* Male 3286 (76.5) 2336 (77.6) 950 (73.9) Female 1010 (23.5) 675 (22.4) 335 (26.1) Type of forensic case, n (%) c< 0.001* Forensic examination (Before or after detention) 1468 (34.2) 1062 (35.3) 406 (31.6) 0.02* Suicide 119 (2.8) 73 (2.4) 46 (3.6) 0.035* Work accident 770 (17.9) 540 (17.9) 230 (17.9) 0.978 Vehicle traffic accident 296 (6.9) 226 (7.5) 70 (5.4) 0.015* Pedestrian accident 141 (3.3) 113 (3.8) 28 (2.2) 0.008* Motorcycle accident 252 (5.9) 157 (5.2) 95 (7.4) 0.005* Assault 1057 (24.6) 680 (22.6) 378 (29.4) 0.001* Stabbing 80 (1.9) 54 (1.8) 26 (2) 0.610 Gunshot wound 13 (0.3) 8 (0.3) 5 (0.4) 0.547 Falling from high 80 (2.7) 1 (0.1) 81 (1.9) <0.001* Burn 15 (0.5) 0 (0) 15 (0.3) Electrical burn 1 (0.1) 0 (0) 1 (0.1) Drowning 2 (0.1) 0 (0) 2 (0.1) *p < 0.05 aPearson chi-square test bFisher-Freeman-Halton exact test cFisher’s exact test Table 2 Distribution of forensic cases by gender before and during the pandemic Female Male Pre-pandemic Time cases Pandemic Time cases p Pre-pandemic Time cases Pandemic Time cases p (n = 675) (n = 335) (n = 2336) (n = 950) Forensic case, n (%) Forensic examination ( before or after detention) 162 (24) 92 (27.5) a0.232 900 (38.5) 314 (33.1) a0.003* Suicide 38 (5.6) 29 (8.7) a0.069 35 (1.5) 17 (1.8) a0.544 Work accident 101 (15) 34 (10.1) a0.034* 439 (18.8) 196 (20.6) a0.226 Vehicle traffic accident 89 (13.2) 14 (4.2) a< 0.001* 137 (5.9) 56 (5.9) a0.974 Pedestrian accident 46 (6.8) 7 (2.1) a0.002* 67 (2.9) 21 (2.2) a0.290 Motorcycle accident 21 (3.1) 7 (2.1) a0.352 136 (5.8) 88 (9.3) a< 0.001* Assault 179 (26.5) 148 (44.2) a< 0.001* 501 (21.4) 230 (24.2) a0.084 Stabbing 8 (1.2) 4 (1.2) 46 (2) 22 (2.3) Gunshot wound 0 (0) 0 (0) 8 (0.3) 5 (0.5) Falling from high 24 (3.6) 1 (0.1) 0 56(2.4) 1(0.1) Burn 6(0.9) 0 (0) 9(0.4) 15 (0.5) 0 Electrical burn 1 (0.1) 0 (0) 0 1 (0.1) 0 Drowning 2 (0.1) 0 (0) 2 (0.1) 2 (0.1) 0 *p < 0.05 aPearson chi-square test Discussion COVID-19 started in China in December 2019 and was acknowledged as a pandemic by the World Health Organization (WHO) in January 2020. Due to the pandemic, we are going through a period in which both our physical and mental health are affected by the national economies, health systems, and individuals. Restrictions to prevent the spread of the disease have resulted in social isolation. The closure of non-essential businesses has also resulted in financial difficulties. These negative factors may lead to the worldwide increase of loneliness, anxiety, hopelessness, suicidal tendencies, and domestic violence [4–6, 10]. In addition to these issues, a decrease in traffic accidents was expected as a result of the decrease in the number of pedestrians and vehicles on roads due to the measures taken [7–9, 11]. Events that follow judicial instructions and cause the deterioration of people’s physical and mental health through external factors, such as traffic accidents, suicide attempts, firearm or penetrating instrument injuries, physical or sexual violence, work accidents, and poisoning, are defined as judicial cases. Forensic cases are usually first admitted to emergency services, and their diagnosis and medical intervention are mostly provided by emergency services [1, 12, 13]. Various studies in our country have reported the mean age of forensic cases as between 27 and 33 years. In our study, the mean age is 31.70 ± 12.64 years, which is similar to that reported in the literature, and there is no difference in terms of the mean age between the pre-pandemic period and the pandemic period [13, 14]. In this study, we found a significant difference in terms of gender for both the pre-pandemic and pandemic periods. Men were more likely to be admitted to the emergency department for forensic reasons than women in both groups. We found that the rate of women’s admissions during the pandemic period was significantly higher than in the pre-pandemic period. In studies conducted before the pandemic, work accidents and traffic accidents were reported more frequently for men in our country, and it was observed that the majority of forensic cases involved men [13–15]. This may be because men are more involved in social and business life [16, 17]. This study indicates a statistically significant increase in the percentage of admissions due to assault on women during the pandemic period compared to the pre-pandemic period. It has been reported that domestic violence increased at different rates in different countries during the pandemic [10, 18]. Domestic violence is generally defined as the physical, emotional, economic, or sexual abuse of the weak by the strong. It can refer to violence of partners against each other, or it can refer to violence against a child or elderly person at home [10, 12, 19]. Every member of society can experience domestic violence, but it has been reported in the literature that women are exposed to higher rates of violence compared to men [10, 12, 19]. Stress factors, unemployment, decreased income, decreased social support, and alcohol and substance use are among the factors that cause domestic violence [6, 10, 12, 19–22]. In addition, it has been demonstrated that domestic violence increases during natural or man-made disasters [6, 18, 22, 23]. In this study, the percentage of admissions due to suicide attempts increased significantly during the pandemic period compared to the pre-pandemic period. When we look at gender, we note that this increase is reported in the female group. Due to the negative effects of the pandemic, there may be an increase in suicidal attempts, and similar results have been reported in the literature [24–27]. Buschmann and Tsokos evaluated 11 suicide cases following the restrictions due to the COVID pandemic. They stated that all the patients had underlying psychiatric conditions, that their COVID tests had been negative, and that they had high levels of anxiety according to the anamnesis taken from the relatives of the patients [27]. Existing psychiatric disorders may worsen with social isolation, and depressive disorders and suicidal tendencies may increase when social support is removed [25–27]. The fact that women have less economic and social support, or limited access to existing support due to the pandemic, and an increase in domestic violence may also be reasons for the higher incidence of suicide among women. Since our study was retrospective, we cannot say exactly how many of the women had experienced domestic violence. However, according to social facts, it is possible to say that these assault cases occurred mostly inside the home. Countries have implemented different degrees of restriction to prevent the spread of the disease during the pandemic. Schools, non-essential workplaces, and entertainment venues were closed, travel restrictions were imposed, and a curfew was imposed in our country. Due to the effect of these restrictions, the number of pedestrians and vehicles on the road decreased. Similar effects were observed globally [8, 9, 11, 28]. The expected effect of the reduction in traffic was a reduction in traffic and vehicle-related accidents. In some countries, such as the USA, Australia, England, Spain, and Denmark, it was observed that traffic accidents decreased in line with this expectation [8, 9, 11]. By contrast, Hakkenbrak et al. reported an increased number of traffic-related accidents in the Netherlands; this may be because there was no stay-at-home order or curfew in the Netherlands [29]. On the other hand, Tandon et al. observed that, although traffic decreased in the US state of Virginia, there was no decrease in traffic-related accidents; this may be due to a decrease in driving safety due to empty roads and an increase in alcohol use [24]. In our study, a decrease was observed in the number of trauma cases due to in-vehicle and pedestrian traffic accidents during the pandemic period. However, the increase in motorcycle accidents during the pandemic period was statistically significant (p = 0.005). These results support the expectation that, with the decrease in traffic, pedestrian and vehicle accidents will also decrease. During the pandemic in our country, the curfew and working from home increased online shopping. Therefore, we believe that the percentage of motorcycle accidents among men increased due to the increase in motorcycle couriers, with the majority of these couriers being men. Yasin et al. reported that, while the number of pedestrian and motor vehicle accidents decreased in the UAE, the number of motorcycle accidents increased [30]. Conclusion This study reveals that the pandemic also affected the number of patients who were admitted to the emergency department for forensic reasons. According to our results, the percentage of suicide attempts, assault cases among women, and motorcycle accidents among men increased during the COVID-19 pandemic. We believe that fear of the disease due to the pandemic, the losses it caused, and the hopelessness and uncertainty about the future people experienced caused an increase in judicial incidents such as suicide and violence. Social support programs involving state and non-governmental organizations can be provided to enable people to stay in touch with each other, and perhaps such forensic incidents can be reduced by providing remote access to the health system, by developing applications such as telemedicine. Limitations Our study represents a retrospective single-center study. Our hospital is located in a large metropolis and our city hosts a wide variety of people in terms of culture and faith. However, forensic cases can be affected by geographical and cultural factors. The results of this study can be generalized to a multicenter study and prospective studies should involve different regions in terms of geography, culture, and religious beliefs. This study was conducted after the first COVID-19 case was announced in our country. Different results may emerge from studies covering longer periods due to fluctuations in people’s moods. Another limitation of the study is that we cannot say exactly how many battered women have been exposed to domestic violence. In future studies, it would be appropriate to examine children and women’s exposure to domestic violence. Since our study is a retrospective study, cross-cutting groups such as suicidal traffic accidents could not be distinguished. This is an important limitation of our study. Key points The COVID-19 pandemic also affected forensic cases: The rate of suicide attempts increased compared to the pre-pandemic period. Domestic violence against women increased. The rate of motorcycle accidents increased. Author contribution FSD designed and carried out the research, coordinated the study, and participated in all of the research.  FSD and TCÖ collected the data and prepared the manuscript.  FSD and TCÖ assisted in designing and conducting the research. FSD and TCÖ participated in manuscript preparation and performed the statistical analysis. FSD and TCÖ corrected the English manuscript and revised further statistical data. All authors have read and approved the content of the manuscript. Declarations Conflict of interest The authors declare no competing interests. Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References 1. Aktas N Gulacti U Lok U Characteristics of the traumatic forensic cases admitted to Emergency Department and errors in the forensic report writing Bull Emerg Trauma 2018 6 1 64 70 10.29252/beat-060110 29379812 2. Sari Dogan F, Guneysel O, Ozaydin V, et al. Effects of Ramadan on forensic cases presenting to Emergency Service. JMSR. 2015;1(4):106–109. 3. Seviner M Kozacı N Ay M O Analysis of judicial cases at Emergency Department Cukurova Med J 2013 38 2 250 60 4. Mota P Avoiding a new epidemic during a pandemic: the importance of assessing the risk of substance use disorders in the COVID-19 era Psychiatry Res 2020 290 113142 10.1016/j.psychres.2020.113142 32502828 5. Holmes EA, O’Connor RC, Perry H, et al. Multidisciplinary research priorities for the COVID-19 pandemic: a call for action for mental health science. Lancet Psychiatry. 2020;7(6):547–560. 10.1016/S2215-0366(20)30168-1. Epub 2020 Apr 15. PMID: 32304649; PMCID: PMC7159850. 6. Telles LEB Valenca AM Barros AJS Domestic violence in the COVID-19 pandemic: a forensic psychiatric perspective Braz J Psychiatry 2021 43 233 4 10.1590/1516-4446-2020-1060 32491034 7. Oguzoglu U. COVID-19 lockdowns and decline in traffic-related deaths and injuries. IZA DP No. 13278. May 2020. 8. Yasin YJ, Grivna M, Abu-Zidan FM. Global impact of COVID-19 pandemic on road traffic collisions. World J Emerg Surg. 2021;28:16(1):51. 10.1186/s13017-021-00395-8. PMID: 34583713; PMCID: PMC8478263. 9. Nuñez JH, Sallent A, Lakhani K, et al. Impact of the COVID-19 pandemic on an emergency Traumatology Service: experience at a Tertiary Trauma Centre in Spain. Injury. 2020;51(7):1414–1418. Published online 2020 May 13. 10.1016/j.injury.2020.05.016. 10. Rosida L, Putri IM, Komarudin K, et al. The domestic violence during the COVID-19 pandemic: scoping review. Open Access Maced J Med Sci. 2021;9(F):660–667. 10.3889/oamjms.2021.7378. 11. Trier F Fjølner J Raabe N Effect of the COVID-19 pandemic at a major danish trauma center in 2020 compared with 2018– 2019: a retrospective cohort study Acta Anaesthesiol Scand 2022 66 265 72 10.1111/aas.13997 34748218 12. Huecker MR Smock W Domestic violence. StatPearls 2020 Treasure Island (FL) StatPearls Publishing 13. Akgün FS Evaluation of the admitted forensic cases to the emergency department Med Sci 2019 8 1 166 8 14. Korkmaz T Kahramansoy N Erkol Z Acil Servise Başvuran Adli Olguların ve Düzenlenen Adli Raporların Değerlendirilmesi Haseki Tıp Bülteni 2012 50 14 20 15. Bilgin EU Meral O Kocak A 2011 yılında trafik kazası sonucu Ege Üniversitesi Hastanesi Acil Servisine başvuran hastaların adli tıbbi boyutuyla incelenmesi Ege Tıp Derg 2013 52 2 93 9 16. https://data.tuik.gov.tr/Bulten/Index?p=Istatistiklerle-Kadin-2020-37221. Accessed 17 Feb 2022. 17. Dirik N. Türkiye’ de kadın istihdamının önündeki engellerin aşılmasında girişimciliğin önemi ve kamu istihdam kurumların rolü. Uzmanlık tezi. Ankara. 2016 https://media.iskur.gov.tr/15674/nuriye-dirik.pdf. 18. Moawad AM Desouky EED Salem MR Violence and sociodemographic related factors among a sample of egyptian women during the COVID-19 pandemic Egypt J Forensic Sci 2021 11 29 10.1186/s41935-021-00243-5 34691785 19. Weil A. https://www.uptodate.com/contents/intimate-partner-violence-diagnosis-and-screening. Accessed 16 Feb 2022. 20. Lyons B, Francys C. Martin CF, et al. Screening, detection, and intervention for domestic violence / Intimate Partner Violence. https://cdn.ymaws.com/www.dcmsonline.org/resource/resmgr/files/nefm/view_pdf/domestic_violence_final.pdf. Accessed 16 Feb 2022. 21. Andrew M Campbell An increasing risk of family violence during the Covid-19 pandemic: strengthening community collaborations to save lives Forensic Sci International: Rep Volume 2020 2 100089 22. Viero A Barbarab G Montisci M Violence against women in the Covid-19 pandemic: a review of the literature and a call for shared strategies to tackle health and social emergencies Forensic Sci Int 2021 319 110650 10.1016/j.forsciint.2020.110650 33340849 23. First JM, First NL, Houston JB. Intimate partner violence and disasters: a framework for empowering women experiencing violence in disaster settings. Affilia: J Women Soc Work. 2017;32(3) 390–403. 24. Tandon R COVID-19 and suicide: just the facts. Key learnings and guidance for action Asian J Psychiatr 2021 60 102695 10.1016/j.ajp.2021.102695 33992557 25. Sher L. The impact of the COVID-19 pandemic on suicide rates. QJM. 2020;113(10):707–712. 10.1093/qjmed/hcaa202. PMID: 32539153; PMCID: PMC7313777. 26. Banerjee D Kosagisharaf JR Sathyanarayana Rao TS ‘The dual pandemic’ of suicide and COVID-19: a biopsychosocial narrative of risks and prevention Psychiatry Res 2021 295 113577 10.1016/j.psychres.2020.113577 33229123 27. Buschmann C Tsokos M Corona-associated suicide - observations made in the autopsy room Legal Med (Tokyo) 2020 46 101723 10.1016/j.legalmed.2020.101723 28. Leichtle SW Rodas EB Procter L The influence of a statewide “Stay-at-Home” order on trauma volume and patterns at a level 1 trauma center in the united states Injury 2020 51 11 2437 41 10.1016/j.injury.2020.08.014 32798035 29. Hakkenbrak NAG, Loggers SAI, Lubbers E, et al. COVID-trauma collaborator group. Trauma care during the COVID-19 pandemic in the Netherlands: a level 1 trauma multicenter cohort study. Scand J Trauma Resusc Emerg Med. 2021;29(1):130. 10.1186/s13049-021-00942-x. PMID: 34493310; PMCID: PMC8423597. 30. Yasin YJ, Alao DO, Grivna M, et al. Impact of the COVID-19 pandemic on road traffic collision injury patterns and severity in Al-Ain City, United Arab Emirates. World J Emerg Surg. 2021;16(1):57. 10.1186/s13017-021-00401-z. PMID: 34798873; PMCID: PMC8602977.

36520378

PMC9753859

NO-CC CODE

2022-12-16 23:26:35

no

Forensic Sci Med Pathol. 2022 Dec 15;:1-6

utf-8

Forensic Sci Med Pathol

10.1007/s12024-022-00565-3

oa_other

==== Front Empirica (Dordr) Empirica (Dordr) Empirica 0340-8744 1573-6911 Springer US New York 9556 10.1007/s10663-022-09556-7 Original Paper Beyond the Covid-19 pandemic: remote learning and education inequalities http://orcid.org/0000-0002-6239-7539 Bonacini Luca l.bonacini@unibo.it 1 http://orcid.org/0000-0002-5212-8626 Murat Marina marina.murat@unimore.it 2 1 grid.6292.f 0000 0004 1757 1758 Department of Economics, GLO, University of Bologna, Bologna, Italy 2 grid.7548.e 0000000121697570 Department of Economics Marco Biagi, IEI, GLO, University of Modena and Reggio Emilia, Modena, Italy Responsible Editor: Harald Oberhofer. 15 12 2022 130 18 11 2022 © The Author(s), under exclusive licence to Springer Science+Business Media, LLC, part of Springer Nature 2022, Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. Is remote learning associated with education inequalities? We use PISA 2018 data from five European countries—France, Germany, Italy, Spain and the United Kingdom—to investigate whether education outcomes are related to the possession of the resources needed for distance learning. After controlling for a wide set of covariates, fixed effects, different specifications and testing the stability of coefficients, we find that remote learning is positively associated with average education outcomes, but also with strong and significant education inequalities. Our results show that negative gaps are larger where online schooling is more widespread, across countries, locations, and school types. More generally, remote learning inequalities appear to be associated with technological network externalities: they increase as digital education spreads. Policy makers must guarantee to all students and schools the possession of the resources needed for remote learning, but to reach this goal efficiently they must adapt their actions to the characteristics of countries, areas and school systems. Supplementary Information The online version contains supplementary material available at 10.1007/s10663-022-09556-7. Keywords Education inequalities Remote learning Technological networks PISA Covid-19 JEL Classification I21 I24 H52 ==== Body pmcIntroduction Several developed countries had already adopted some form of remote learning when the Covid-19 pandemic struck in 2020 and, abruptly, turned it into the main form of schooling. Many governments reacted rapidly, providing various degrees of support to schools and students lacking the resources needed to online schooling, but the urgency of these actions lead to uneven results across areas, schools and students’ populations. With the ebb of the pandemic and the reopening of schools, the sense of urgency subsided and government interventions slowed. On the research side, the pandemic gave rise to many studies on the effects of remote learning on education. However, most of them restrict the analysis to online education during school closures, without extending its reach beyond pandemic times. A different branch of research on the use of ICT tools in schooling is based on experimental or quasi-experimental tests, but also in this case the investigation does not produce results that can be generalized to the impact of remote and online learning on education. The present research investigates whether an uneven possession of the resources needed for remote learning by students and schools is associated with inequalities in education. It differs from the studies inspired by school closures during the Covid-19 pandemic (Kuhfeld et al. 2020; Schleicher 2020) as well as from the experimental and quasi-experimental research measuring the performance students—often from disadvantaged socioeconomic backgrounds or in developing economies—who are provided with computers or internet access. Overall, the results of these studies are heterogeneous, with computer and internet in some cases leading to improved and in other cases to poorer education outcomes (Banerjee et al. 2007; Leuven et al. 2007; Fairlie 2012; Carter et al. 2017; Comi et al. 2017; Malamud et al. 2019). Instead, we focus on remote learning and students’ education outcomes during non-critical times and use a very comprehensive and large dataset that covers a relatively homogenous group of developed countries. More specifically, we use the 2018 wave of the Program for International Student Assessment (PISA) survey to assess whether the educational outcomes of students unable to learn remotely are significantly different from those of their peers. The PISA dataset provides comparable data within and across countries and is representative of countries’ entire students’ populations of 15-year-olds. It measures students’ ability to use their reading, mathematics and science knowledge and skills every three years and is characterized by features of standardization and comprehensiveness that allow the implementation of cross-country comparisons over several dimensions, including the one that motivates this study, remote learning. To our knowledge, this is the first time that this type of research is based on such a wide dataset. We focus on the scores in mathematics and consider countries—France, Germany, Italy, Spain and the United Kingdom—that share the economic, institutional and social characteristics of the western European area, while at the same time partially differ in their educational systems. This allows results to be independent from structural differences in the level of development of countries, but at the same time lets them vary with schooling systems. Our variables of interest are, at home, the availability of a computer for schoolwork, an internet connection and a quiet place to study, and, at school, of a platform for online schooling. Since the possession of these resources can be expected to be correlated with the characteristics of students and of their families and schools, we control for a wide array of potential confounders, fixed effects and specifications. In fully controlled regressions, we find negative strong and significant gaps in the education of students unable to learn remotely. Perhaps unexpectedly, they are larger in the United Kingdom, where students and schools are best endowed with the resources needed for digital learning, and smaller in Spain, where digital education is less widespread. Specifically, the lack of a computer for schoolwork is associated with negative gaps in mathematics that range from the equivalent to half of a school year in the United Kingdom to about a sixth of a year in Spain. Moreover, gaps tend to be larger in urban areas, where the use of ICT resources is more common, and in the best endowed schools. In general, our results suggest that there are network technological externalities in remote learning that make the losses of outsiders larger where online learning is more widespread. We also find that a quiet place to study is significantly related to students’ scores. Composition analyses show that gap variations are partly explained by school types in countries where school tracking starts earlier, by grade repetition where repeating grades is more frequent, and by socioeconomic factors in other cases. In countries where grade repetition is frequent, remote learning inequalities are also associated with students’ joint probabilities of repeating grades and planning to abandon education early. Our results evidence correlations rather than causal relationships, but our sets of controls on socioeconomic factors, individual characteristics, school systems and fixed effects contribute to make them quite robust. While several covariates help to explain part of the variations in remote learning gaps, these inequalities tend to remain strong and significant even in the fully controlled regressions, which points to a direct relationship between the scarcity of the resources needed to remote learning and school outcomes. Hence, our results signal the necessity of decisive policy actions even in non-pandemic times, aimed at ensuring that all students and schools possess the resources needed for online learning. The findings on the correlates of these gaps show how these actions can be tailored according to the characteristics of countries, areas and school systems, but the fact that gaps can remain strong and significant even after all correlates have been accounted for, indicates that these actions must also be quite direct. The rest of the paper is structured as follows: Sect. 2 discusses the related literature, Sect. 3 presents the data and main descriptive statistics, Sect. 4 is dedicated to the empirical methodology, results are provided in Sects. 5, and 6 concludes. Related literature The impact of digital and remote schooling on students’ outcomes has been widely debated for at least the last two decades. A group of publications frequently cited is based on a randomized control experiment performed in 2006. It consisted into a random assignment to first-year community college students in California of computers to be used at home, and aimed to estimate educational outcomes and labour market returns. Since the goal was the evaluation of the effects of home computers alone, no training, other assistance or resources were provided. Fairlie (2012) finds that the treatment group who received home computers developed substantially better computer skills than the control group, and Fairlie and London (2012) that the treated students experienced small, positive, short-run effects on educational outcomes. However, Fairlie and Robinson (2013) show that, although computer use increased substantially among the treated, there were no effects on educational outcomes, including grades, standardized test scores, or others. Fairlie and Grunberg (2014) evidence that the treatment group of students had a higher probability of taking transfer courses—allowing them to move from community college to university—than the control group. Finally, Fairlie and Bahr (2018) examine the short- to medium-term effects on earnings, employment and college enrolment, without finding significant effects of computer skills on college enrolment or short- or medium-run earnings. Among studies focusing on developing countries, Banerjee et al. (2007) analyse the impact of two different programs, implemented in 1998 and in 2000, that provided supplementary inputs to children from poor families in urban India: a remedial education and a computer-assisted learning program. The second intervention offered children in grade four two hours of shared computer time per week during which they played games that involved solving math problems. Results show that both programs had a substantial positive effect on children’s academic achievement, at least in the short run. Malamud et al. (2019) study the effects of home internet access by considering a broad range of child outcomes in Peru during the years 2011 to 2013. Data derive from an experiment consisting into randomly providing low-cost laptops (XO) for home use to children enrolled in low-achieving public primary schools, and into selecting among them a subgroup to whom also supply a free high-speed internet access. They find that children with internet access improved their computer and internet proficiency relative to those without computers, and improved their internet proficiency relative to those with computers only. However, there were no significant effects of internet access on math and reading achievements, cognitive skills, self-esteem, teacher perceptions, or school grades when compared to either group. Beuermann et al. (2015) focus on the short-term effects—approximately five months after performing the experiment—of providing children with XO laptops but not internet access. Results are that scores in an objective test measuring proficiency in using the XO laptop increased, but math and reading scores did not significantly change. Malamud and Pop-Eleches (2011) estimate the effect of home computers on child and adolescent outcomes in Romania by exploiting a voucher program, subsidized in 2008 by the Romanian Ministry of Education, which awarded approximately 35,000 vouchers worth 200 Euros towards the purchase of a personal computer to low-income students enrolled in public schools. They employ a regression discontinuity design that allows comparisons across students who are very similar in family income and other respects but markedly differ in their possession of a computer at home. Their results indicate both positive and negative effects of home computers on the human capital development: children who won a voucher to purchase a computer had significantly lower school grades but showed improved computer skills. Among researchers considering more developed economies, Leuven et al. (2007) evaluate the effects of two types of subsidies provided in Netherlands in 2000 to schools where large proportions of students had parents with low levels of education. The first scheme provided extra funding for personnel, the second for computers and software. To identify the effect of the two programs on pupils’ achievement, they exploit regression discontinuities in a local difference-in-differences framework. They find that the effects of both types of subsidies are negative and, in some cases, significant. Moreover, computer subsidies worsened girls’ achievements. Carter et al. (2017) analyse the short-term effects of using ICT tools at school by employing data from an experiment performed in 2015 in the United States Military Academy, a four-year undergraduate institution. By considering final exam scores as the outcome variables, they randomized classrooms into a control group, where students were not allowed to use laptops or tablets, and two treatment groups, one where students were allowed to freely use them during class for note-taking and classroom participation, the other where students could use them, but could be monitored by lecturers. They find negative effects in both types of treatment, suggesting that using these ICT devices at school can harm classroom performance even when their utilization is monitored. A few studies are based on wider samples. One is Yanguas (2020), who examines the early-adulthood educational outcomes of students who were provided laptops and internet access as school children in 2007 during the Plan Ceibal in Uruguay. This nationwide one-laptop-per-child program delivered a laptop to each student in primary and middle schools within the public education system and equipped all public schools with wireless internet access. She finds negative effects of the program on educational attainment. Students who were exposed to the program were less likely to apply for scholarships and to enrol in technology-related majors relative to health and social sciences majors.1 Vigdor et al. (2014) rely on a longitudinal sample consisting on students enrolled in grades five through eight in public schools in North Carolina between the years 2000 and 2005. They address concerns of non-random selection by employing a within-student estimator and using local variation in the timing of the introduction of broadband internet services; then they trace the impact of home computer introduction for a period of up to three years. They document broad racial and socioeconomic gaps in home computer access and use, and find that the introduction of home computer technology is associated with modest, but statistically significant and persistently negative relationships with students’ math and reading test scores. Moreover, the introduction of broadband internet is associated with widening racial and socioeconomic achievement gaps. They speculate that broadband internet access can reduce the efficiency of the time spent on homework, presumably by introducing distractions and new options for leisure time. Using longitudinal data from years 2001 to 2006 and a difference-in-difference approach Cristia et al. (2014) find that the introduction of computers in a school in Peru have no significant effects on students’ repetition, dropout and enrolment rates. Comi et al. (2017) analyse whether ICT-related teaching practices affect students’ achievement in the Lombardy region in Italy by using standardized survey data (INVALSI) and an ad-hoc ICT survey performed in 2012 on a representative sample of students and teachers from 100 classrooms in the second year of upper secondary school (10th grade). To address issues of endogeneity, they use a within-student between-subject estimator, which controls for unobserved heterogeneity in schools, classrooms and students. They find that computer-based teaching practices increase student performance if they are aimed at increasing students’ awareness of ICT use and at improving their navigation critical skills, ability to distinguish between relevant and irrelevant material and access, locate, extract, evaluate and organize digital information. On the other hand, they report a negative impact of practices requiring an active role of students in classes using ICT. Hence, the results of the above studies are heterogeneous. Several of them are based on experimental or quasi-experimental approaches that focus on samples of students from disadvantaged backgrounds, or from less well-off areas or countries where the use of ICT technologies for schoolwork is limited. Moreover, in some experiments students are supplied with ICT tools without being provided with previous training, which can affect results. In some cases, regarding developed economies, the experiments have been performed at times when online learning was still uncommon, so that results can hardly apply to present times. A few studies are based on broad samples, but results between them also diverge. Overall, the findings on the effects of the use of ICT tools in education differ strongly, partly because of the heterogeneity of the samples, methodologies and periods of time considered, and partly because the underlying research questions also vary. Several research approaches and findings on this topic are excellently reviewed in Escueta et al. (2020). Data and descriptive statistics We use data from the 2018 wave of PISA assessment on students’ test scores in mathematics.2 Overall, we consider 73,305 students enrolled in over 2577 schools in the five countries. The PISA dataset is the result of a two-stage stratified design, where, first, individual schools are sampled, and secondly, students are randomly sampled within schools. Given that each participating student in the PISA survey answers a limited amount of questions taken from the total test item pool, OECD provides ten test scores (known as plausible values), which can be interpreted as multiple imputed values of students’ performance based on students’ answers to the test and their background questionnaires. The difficulty of each item represents a weight, used to compute the weighted averages of correct responses. This approach allows having a measure of an individual’s proficiency for each student in each subject area, regardless of the questions actually answered. We employ the recommended OECD strategy for the estimation of coefficients and their variances, making use of all ten plausible values (OECD 2018). As a result, the number of students in the nationally defined target populations represented by our analytical samples covers from 85% (United Kingdom and Italy) to 99% (Germany) of the five countries’ populations of 15-year-olds (more details are in the Online Appendix). Regarding the effective possibility of learning remotely, we select from the PISA Student’s Questionnaire the following questions: Which of the following are in your home: A computer you can use for schoolwork, A quiet place to study, A link to the internet, with responses that can be ‘Yes’ or No’, and from the School’s Questionnaire: To what extent do you agree with the following statements about your school’s capacity to enhance learning and teaching using digital devices? An effective online learning support platform is available, with answers that vary from ‘Strongly disagree’ to ‘Strongly agree’.3 Concerning the planned length of students’ education, from the Students’ Questionnaire we consider: Which of the following do you expect to complete? Answers range from lower secondary to advanced tertiary and research education programs. Our main control variables concern student’s individual characteristics (gender, immigrant status, age at arrival if born abroad, and repetition of one or more school years), family’s socioeconomic status (parents’ education and occupation, number of books and e-books at home), type of school attended (general, technical or professional, and private or public) and its location (city, town or rural). A detailed definition of variables is in Table 6. In further specifications we add school fixed effects. Descriptive statistics are summarised in Table 5 and correlations between our main variables are in Table S1. The latter shows, in particular, that there is a weak correlation between our four variables of interest. It also shows that the less abundant resource is the ICT endowment at school: its scarcity is highest in Germany and lowest in the United Kingdom. If only home ICT resources—computer for schoolwork, an internet connection—are considered, the country with the highest proportions of students lacking them is Italy and the one with the lowest scarcity is the United Kingdom. The latter, however, has the highest proportion of students without a quiet place to study at home. If the proportion of students lacking at least one of the four resources needed for remote learning is considered—a computer, an internet connection, a quiet place to study at home, a school with a platform for online teaching—Figure S1-A evidences that it varies from 50% in the United Kingdom (42% if a quiet place to study is excluded) to 74% in Germany (70% considering only ICT resources). Grade repetition is unusual in the United Kingdom but common in the other four countries, especially Spain and Germany, where 29 and 20% of students repeat grades, respectively. Educational systems also differ in the degree of tracking between schools: the age at which students are tracked for the first time is 10 years in Germany, 14 in Italy, 15 in France and 16 in the United Kingdom and Spain (in the latter, however, some of vocational schools start at 15; Woessmann 2009). The proportion of students planning to leave education early varies from about 30% in Germany (where vocational school can be attended while working part-time) to six percent in Italy. Since secondary studies can be completed at a different age in each of the five countries—compulsory education ends at the ages of 18 or 19 in Germany (depending on each länder) and at 16 in the other four countries—the definition of what is early varies with each institutional setting. Empirical strategy To gauge the links between remote learning and education outcomes, we test, separately for each country, the relationships between the students’ scores in mathematics and the lack of the resources needed to learn remotely by using the following specification:1 Testscoresij=α1+β1Nocomputerij+β2Nointernetij+β3Noquietplaceij+β4NoschoolICTj+XijΠ+ZijΛ+SijΓ+λj+εij where Test score is the weighted test score in mathematics of student i in school j, No computer, No internet, No quiet place, No school ICT are the variables of interest. Xij, Zij and Sij are sets of covariates concerning, respectively, student’s characteristics, family’s socio-economic status, and the type of school attended. Specifically, the vector of student’s characteristics, Xij, includes Female, a dichotomous variable taking value one if the student is female and zero otherwise, Age, in months, Immigrant status, the student’s status of immigration (a dichotomous variable taking value one if the student is immigrant and zero otherwise), Age at arrival in months (taking value zero if native) and Repeated grade if the student repeated one or more school years; the vector concerning the family’s socio-economic status, Zij, includes Parents’ education (the highest level of education among parents, HISCED in PISA), Parents’ occupation (the highest occupational status among parents, HISEI) and the number of Books and of e-Books at home; Sij are school characteristics, denoting general, technical or professional schools, and private or public schools; λj are school fixed effects, and εij are error terms clustered at the school level.2 Leavingeducationearlyij∗=α1+β1Nocomputerij+β2Nointernetij+β3Noquietplaceij+β4NoschoolICTj+WijΠ+ZijΛ+SijΓ+ε1ij In Sect. 5.5, we use separate Probit specifications to gauge the correlations between the probabilities of leaving education early and of repeating a grade (except for the United Kingdom, where grade repetition is uncommon) with our four variables of interest. The variable on the students’ planned length of investment in education takes value one when it stops at lower secondary studies or at upper secondary levels leading directly to the labour market, and zero otherwise.Subsequently, we use a Bivariate Probit specification to test whether the joint probabilities of planning to leave school early and repeating a school year are correlated with the lack of the resources needed to learn remotely. The Probit and Bivariate Probit specifications on leaving school early and repeating a school year are: 3 Repeatedgradeij∗=α1+β1Nocomputerij+β2Nointernetij+β3Noquietplaceij+β4NoschoolICTj+WijΠ+ZijΛ+SijΓ+ε1ij With planning to leave education early:Leavingeducationearlyij=1ifLeavingeducationearlyij∗>0Leavingeducationearlyij=0ifLeavingeducationearlyij∗≤0 And Repeated grade:Repeatedgradeij=1ifRepeatedgradeij∗>0Repeatedgradeij=0ifRepeatedgradeij∗≤0 The error terms ε1ij and ε2ij are assumed to be independently and identically distributed as bivariate normal. The vector Wij comprises the same covariates on individual characteristics included in Xij except for Repeated grade, which is now one of the two dependent variables. Since these models are nonlinear, we do not include school fixed effects (Cameron and Trivedi 2005). Results Home and school resources for remote learning Table 1 presents the results of estimating Eq. (1) with a separate sample for each country. Coefficients on our variables of interest are the differences between the scores of students unable to learn remotely and those of their peers. They can also be interpreted as proportions of school years by considering that, on average in OECD countries, the cognitive content of one school year corresponds to about 40 score points (on a mean of 500; OECD 2019). Base regressions include only our four variables of interest, No computer, No internet, No quiet place to study and No school ICT. Each column between the base and full regressions comprises also one group of covariates (included one at a time: individual characteristics, socioeconomic factors or school types). The last two columns include, respectively, all controls and all controls plus school fixed effects (detailed results are in Table S2 in the Online Appendix).Table 1 Remote learning resources France Germany (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) Base Individual Socioeconomic School type Full Full—FE Base Individual Socioeconomic School type Full Full—FE No computer  − 62.533***  − 41.282***  − 38.440***  − 28.252***  − 20.631***  − 19.347***  − 71.246***  − 52.432***  − 39.492***  − 57.387***  − 33.163***  − 20.849*** (5.273) (3.825) (5.226) (3.976) (4.159) (3.977) (6.287) (5.702) (6.709) (6.876) (6.499) (5.729) No internet  − 10.569 0.162  − 18.283  − 12.905  − 15.982  − 3.937  − 51.702***  − 38.863***  − 43.308***  − 38.766***  − 28.775***  − 31.554*** (13.691) (11.136) (13.760) (10.833) (11.530) (13.084) (10.515) (10.473) (10.907) (10.967) (10.549) (10.272) No quiet place to study  − 37.640***  − 18.175***  − 18.148***  − 16.236***  − 5.280  − 3.422  − 31.932***  − 15.253**  − 18.847**  − 22.817***  − 8.973  − 1.283 (5.418) (4.978) (4.772) (4.203) (4.453) (4.405) (6.877) (7.288) (7.951) (7.163) (7.643) (7.898) No school ICT  − 4.631 0.610  − 3.544  − 1.488  − 0.833  − 14.058  − 16.183**  − 11.285*  − 10.539  − 12.553** (7.670) (5.812) (5.162) (4.636) (3.463) (9.921) (7.763) (6.791) (7.890) (5.426) Constant 510.761*** 426.139*** 519.638*** 561.847*** 496.900*** 459.339*** 523.869*** 104.506 534.609*** 560.337*** 66.588 15.171 (5.346) (65.001) (6.675) (5.705) (56.620) (54.655) (7.238) (99.103) (6.488) (20.295) (86.303) (81.055) Individual characteristics No Yes No No Yes Yes No Yes No No Yes Yes Parents characteristics No No yes No Yes Yes No No Yes No Yes Yes School characteristics No No No Yes Yes No No No No Yes Yes No School FE No No No No No Yes No No No No No Yes Observations 5341 5277 4889 5341 4852 4852 4077 3986 3616 4049 3544 3570 R2 0.059 0.264 0.265 0.406 0.476 0.528 0.071 0.203 0.239 0.204 0.343 0.52 Italy Spain (13) (14) (15) (16) (17) (18) (19) (20) (21) (22) (23) (24) Base Individual Socioeconomic School type Full Full—FE Base Individual Socioeconomic School type Full Full—FE No computer  − 47.097***  − 36.129***  − 30.398***  − 31.749***  − 20.058***  − 13.635***  − 47.789***  − 15.847***  − 23.467***  − 44.420***  − 6.857*  − 6.085* (6.204) (5.152) (5.570) (4.763) (4.516) (4.186) (3.287) (3.485) (3.508) (3.321) (3.635) (3.604) No internet  − 43.378***  − 36.380***  − 28.543***  − 28.029***  − 4.651  − 20.642**  − 4.538  − 2.960  − 17.365** 4.770 4.300 (10.444) (10.515) (9.711) (9.213) (9.588) (8.729) (8.112) (7.840) (8.241) (8.046) (7.889) (7.925) No quiet place to study  − 15.754***  − 8.063  − 4.177  − 5.443 3.535 1.287  − 8.917**  − 3.355  − 2.024  − 8.221*  − 0.682 0.476 (5.554) (5.678) (5.768) (5.236) (5.575) (5.054) (4.176) (4.028) (4.329) (4.212) (4.154) (4.005) No school ICT  − 3.507  − 2.991  − 6.76  − 12.906*  − 10.031  − 5.846**  − 2.112  − 2.503  − 3.897  − 0.456 (9.568) (8.885) (7.498) (6.866) (6.130) (2.818) (2.280) (2.006) (2.575) (1.853) Constant 497.262*** 296.172*** 514.277*** 554.359*** 431.558*** 357.889*** 489.972*** 358.087*** 507.958*** 505.281*** 364.337*** 366.909*** (6.690) (76.540) (6.688) (19.812) (70.823) (63.723) (2.049) (40.661) (3.554) (2.848) (41.464) (43.433) Individual characteristics No Yes No No Yes Yes No Yes No No Yes Yes Parents characteristics No No Yes No Yes Yes No No Yes No Yes Yes School characteristics No No No Yes Yes No No No No Yes Yes No School FE No No No No no Yes No No No No No Yes Observations 10,979 10,806 10,386 10,979 10,256 10,256 34,033 33,415 32,182 33,958 31,611 31,680 R2 0.039 0.13 0.154 0.219 0.298 0.524 0.03 0.286 0.174 0.055 0.33 0.395 UK (25) (26) (27) (28) (29) (30) Base Individual Socioeconomic School type Full Full—FE No computer  − 43.098***  − 43.256***  − 22.015***  − 42.167***  − 21.913***  − 22.253*** (5.232) (5.089) (5.303) (5.472) (5.506) (5.604) No internet  − 93.290***  − 76.370***  − 67.620***  − 93.002***  − 70.242***  − 64.219*** (15.029) (12.407) (15.808) (15.088) (17.348) (15.487) No quiet place to study  − 23.206***  − 21.763***  − 10.029**  − 23.419***  − 9.151*  − 8.057* (5.069) (4.880) (4.573) (5.034) (4.768) (4.620) No school ICT  − 18.617**  − 17.795**  − 12.442**  − 16.733**  − 11.229** (7.504) (7.260) (5.117) (7.391) (5.129) Constant 519.002*** 145.815 530.446*** 526.627*** 232.335** 285.673*** (4.090) (111.011) (4.574) (4.840) (105.235) (90.028) Individual characteristics No Yes No No Yes Yes Parents characteristics No No Yes No Yes Yes School characteristics No No No Yes Yes No School FE No No No No No Yes Observations 10,728 10,376 9170 10,699 8930 8954 R2 0.052 0.072 0.169 0.067 0.197 0.321 Dependent variable is student’s scores in mathematics. Standard errors are clustered at the school level. ***p < 0.01, **p < 0.05, *p < 0.1. All plausible values employed. All results are weighted and replication weights are taken into account. Covariates are: gender, age in months, repeated grade, immigrant status, age of arrival as individual characteristics; highest parents’ level of education, highest parents’ level of employment, books at home, e-books at home as parents characteristics; school characteristics are technical, vocational, lyceums; public or private Table 1 shows that the first of our variables of interest, not having a computer at home for schoolwork, is negatively and significantly associated with students’ scores in all countries and specifications. In fully controlled regressions, negative coefficients equal the loss of about half a school year in France, Germany and the United Kingdom, and a third and a sixth of a school year, respectively, in Italy and Spain (or, respectively, of 19.3, 20.8 and 22 negative score points). The size and robustness of these coefficients across countries and specifications evidence, symmetrically, that having a computer at home for schoolwork is positively associated with education outcomes. In the full regressions, the coefficients on the variable evidence the gap that remains after the indirect incidence of covariates on scores channelled by the lack of the possession of a computer has been controlled for. To check for the possibility that these strong relationships are still driven by omitted confounding factors concerning, for example, unobserved socio-economic characteristics, we performed the Oster (2019) test statistic, δ, on our variables of interest. The Oster test assesses the coefficient stability and the potential importance of unobserved variables by comparing a regression without controls with the full regression and with a hypothetical regression that includes both the observed and unobserved controls. A value of δ greater than one implies a selection on observed that is at least as important as a selection on unobserved, and indicates a result robust to omitted variable bias. Following the approach suggested by Oster (2019), we assume a maximum obtainable R-squared equal to 1.3 times the R-squared of the full model. Our results on No computer, show that coefficients in Table 1 are robust and unlikely to be confounded by unobserved characteristics. The values of δ are 3.31 for France, 1.94 for Germany, 1.12 for Italy, 6.57 for Spain, and 14.79 for the United Kingdom. Test results on the other three variables of interest are similar. Findings on the lack of an internet connection at home present more variability across countries. Specifically, coefficients are always negative and significant in the United Kingdom: in the fully controlled within-school regression (column 30) the education loss corresponds to more than one and a half school year (64.2 score points). In Germany results are equally robust and the fully controlled gap equals about 75% of a school year. Coefficients are significant in Italy except for the within-school regression (column 18), which indicates that differences between students in the availability of internet connections at home are correlated with the specific schools attended. However, it can also be observed that the type of school attended (column 17), before school fixed effects are included (in column 18), contributes to explain much of the variation in the coefficient with respect to the base regression (column 13). In Spain, coefficients on No internet are significant in the initial specification, but shrink and lose significance as individual characteristics are considered (columns 19 and 20). Among them, having repeated a grade has a strong and significant negative relationship with scores (Table S2). Coefficients have the expected signs in France, but are not significant. The United Kingdom is the only country where a negative association between the lack of a quiet place to study and students’ scores is significant in all specifications. In the fully controlled regression (column 30) it corresponds to a loss in education of about 20% of a school year. In the other four countries, coefficients are significant in the base regression, but shrink and lose significance especially in relation to socioeconomic conditions and school types in France, Germany and Italy, and to grades repetition in Spain.4 Regarding schools, No school ICT is negatively and robustly associated with students’ scores, again, only in the United Kingdom. Coefficients in Germany are significant in the full regression (column 11), but not in all specifications. In both countries, attending a school without a platform for online teaching implies an education loss of almost 30% of a school year (about 12 score points). Coefficients have the expected signs in the other three countries, but their significance is weak or nil. School locations Among the ICT resources needed to learn remotely, those possessed by schools are as crucial as those available in students’ homes but, as evidenced by the descriptive statistics of Table A1 and Figure S1-B, are scarcer. Moreover, as seen above, coefficients on this variable are weakly significant in all countries except the United Kingdom. This low significance can hide heterogeneities at more disaggregated levels concerning, for example, school locations. Among these, since cities are generally better endowed with internet and broadband infrastructures than rural areas, it can be reasonably expected that urban schools make a higher or more efficient use of platforms for remote teaching than rural ones. Also, in urban locations, the networks of students and schools linked through online teaching can be expected to be stronger. If this is so, everything else equal, the education losses of students unable to access online classes can be expected to be bigger in these locations than in rural areas, where remote learning is less widespread. To test this hypothesis, we use the question in the School Questionnaire: Which of the following definitions best describes the community in which your school is located? to build a categorical variable, Location, where rural areas are populated by less than 3000 people, towns by a number between 3000 and 100,000 people  and cities by more than 100,000 people. Then, we interact this variable with No school ICT. Results in Table 2 concerning the fully controlled regressions show that the coefficients on the interactions of No school ICT with Location (rural areas are in the intercept) are strongly negative and significant in Italy and the United Kingdom, and, although to a lesser degree, also in Spain. Specifically, living in a city or town positively affects scores, but attending a city or town school that does possess a platform for teaching online is associated to strong negative score gaps, which correspond to 52.4 and 48.8 negative scores in Italian cities and towns respectively, and to 32 and 36.9 in the United Kingdom. In Italy, these losses are well above a school year and in the United Kingdom to almost a year. This supports our expectation that students attending schools located where the use of digital devices is more common but that lack the resources needed to teach remotely experience larger cognitive losses than those living in rural locations, where digital education networks are less widespread. In France and Germany locations appear to matter less: neither coefficients on the interacted variables, nor on them separately, are significant. These results can be due to a more homogenous distribution of ICT resources between schools across different areas in these countries.Table 2 No school ICT resources and school locations France Germany Italy Spain UK (2) (4) (6) (8) (10) No computer  − 20.743***  − 32.137***  − 20.397***  − 6.928*  − 21.742*** (4.164) (6.555) (4.458) (3.645) (5.522) No internet  − 15.941  − 29.524***  − 18.253* 4.903  − 70.753*** (11.466) (10.690) (9.552) (7.880) (17.262) No quiet place  − 5.774  − 9.581 3.363  − 0.728  − 9.465** (4.384) (7.674) (5.606) (4.142) (4.770) No school ICT  − 9.136  − 74.769 35.809 12.978* 22.130** (17.747) (57.093) (22.317) (7.560) (10.970) (No school ICT)*(Town) 6.745 68.319  − 43.766*  − 12.219  − 36.867*** (18.570) (56.759) (24.135) (7.819) (12.305) (No school ICT)*(City) 14.759 52.215  − 52.455**  − 16.131*  − 31.953** (20.071) (59.479) (24.662) (8.628) (14.185) Town  − 9.278  − 43.755 18.923 0.415 18.175*** (8.861) (33.830) (18.568) (5.621) (6.172) City  − 11.811  − 48.963 26.013 7.799 16.740** (9.810) (36.741) (20.816) (6.085) (7.172) Constant 503.277*** 103.476 411.688*** 360.925*** 216.580** (57.680) (97.967) (71.748) (42.264) (104.961) Covariates Yes Yes Yes Yes Yes Observations 4852 3522 10,256 31,557 8930 R2 0.476 0.351 0.299 0.331 0.199 Dependent variable is student’s scores in mathematics. Standard errors are clustered at the school level. ***p < 0.01, **p < 0.05, *p < 0.1. All plausible values employed. All results are weighted and replication weights are taken into account. The base level of Location is Rural area. Covariates are: gender, age in months, repeated grade, immigrant status, age of arrival, highest parents’ level of education, highest parents’ level of employment, books at home, e-books at home, school types (technical, vocational, lyceums; public or private) Correlates of school fixed effects The lack of significant coefficients on No school ICT in the full regressions of all countries except Germany and the United Kingdom could also, more generally, depend on ICT school resources being in fact proxies of other factors, such as school wealth or school characteristics. For example, if No school ICT were a proxy of school wealth, the coefficient on the variable would include the effects of schools’ infrastructures and economic resources. All these factors, together with other school characteristics, are absorbed into school fixed effects. Hence, to control whether our results on the associations between the variable No school ICT and students’ scores effectively depend on the lack of a school platform for remote teaching, we regress the coefficients of the school fixed effects (estimated from the full fixed-effects regressions of Table 1: columns 6, 12, 18, 24 and 30) on the variable No school ICT, while controlling for the other school characteristics. Coefficients on fixed effects in the full regressions measure each school’s outcome in terms of scores once all characteristics of schools and individuals have been controlled for. Specifically, to take into account school wealth indicators, we consider two questions in the School Questionnaire: Is your school’s capacity to provide instruction hindered by any of the following issues? The first is: A lack of physical infrastructure (e.g. building, grounds, heating/cooling, lighting and acoustic systems), and the second is: Inadequate or poor quality physical infrastructure (e.g. building, grounds, heating/cooling, lighting and acoustic systems). Answers range from ‘Not at all’ to ‘A lot’. With these, we build two binary variables, Lack of infrastructures and Inadequate infrastructures, which take value one when answers are ‘A lot’ or ‘To some extent’ and zero otherwise. Other than these two variables on school infrastructures, we also control for school types and school locations. Results in Table S3 show that, after controlling for school wealth, school types and locations, the relationship between school outcomes and students’ scores are negatively and significantly related to the lack of a platform for remote teaching across all countries. These results further support the finding in Table 1 that No school ICT has a direct relationship with school outcomes and is it not just an indirect indicator of school wealth. Moreover, they show that when platforms are available they are also effectively used; otherwise, their presence or absence would be just an indicator of other factors and, again, would not be directly correlated with school scores. The results of Table 1, plus those on school locations and on the correlates of the coefficients on school fixed effects evidence that gaps associated to the lack of the ICT resources needed for remote schooling, No computer, No internet and No school ICT, are larger where the use of ICT resources is more widespread: among countries, in the United Kingdom, and among locations, in cities and towns. In particular, the lack of a computer for schoolwork is significant in all countries, but losses in education are bigger where their use is more widespread: United Kingdom, Germany and France. Together, these results are consistent with the existence of technological network externalities in education. Gelbach decomposition Results in Table 1 and Table S2 show that coefficients on our four variables of interest vary across specifications as the different groups of covariates are included into the regressions. They also show also that coefficient variations are not driven by the same sets of covariates across countries. For example, when compared with the base regressions, coefficients on the variables of interest in France vary more with school types (Table 1, column 4) than with other covariates, while in the United Kingdom they vary especially with socioeconomic factors. However, while these and other gap variations can be directly observed in Table 1, their relationships with covariates can be precisely computed by using the Gelbach (2016) decomposition method. It shows how much of the variation of coefficients from the base to the full regressions are due to each cofactor, while at the same time, it is independent from the order in which covariates are added to the regressions. Table 3 shows the decomposition of coefficient variations in relation to the three groups of covariates—concerning individual, socioeconomic and school factors—while Table S4 in the Online Appendix presents detailed results on each variable.Table 3 Gelbach decomposition by country ∆ coefficient Individual (%) Socioeconomic (%) School (%) Total explained (1) (2) (3) (4) France No computer  − 5.18*** 11.54  − 14.77*** 32.90  − 24.94*** 55.56  − 44.89*** No internet  − 0.25 6.06 0.6 6.41 No quiet place to study  − 6.21*** 20.53  − 9.06*** 29.95  − 14.98*** 49.52  − 30.25*** No school ICT  − 1.17  − 0.6  − 2.14  − 3.92 Germany No computer  − 7.65*** 20.92  − 20.47*** 55.97  − 8.46*** 23.13  − 36.57*** No internet  − 10.18* 46.89  − 4.98 22.94  − 6.55** 30.17  − 21.71** No quiet place to study  − 9.13*** 31.31  − 13.85*** 47.50  − 6.18** 21.19  − 29.16*** No school ICT 1.06  − 2.87  − 4.16  − 5.98 Italy No computer  − 3.32** 14.41  − 8.18*** 35.50  − 11.54*** 50.09  − 23.04*** No internet  − 3.63 14.94  − 8.34*** 34.32  − 12.34** 50.78  − 24.30*** No quiet place to study  − 3.52** 20.34  − 5.43*** 31.37  − 8.37*** 48.35  − 17.31*** No school ICT  − 2.24* 2.14 7.54* 7.45 Spain No computer  − 26.28*** 62.45  − 14.96*** 35.55  − 0.84** 2.00  − 42.08*** No internet  − 15.83*** 60.08  − 10.11*** 38.37  − 0.42 1.59  − 26.35*** No quiet place to study  − 3.86*** 49.42  − 3.78*** 48.40  − 0.17 2.18  − 7.81*** No school ICT  − 3.67*** 53.34  − 2.60*** 37.79  − 0.62 9.01  − 6.88*** United Kingdom No computer 0.33  − 1.70  − 18.95*** 97.63  − 0.8 4.12  − 19.41*** No internet 3.07  − 13.94**  − 1.7  − 12.58 No quiet place to study  − 1.60* 11.39  − 12.71*** 90.46 0.26  − 1.85  − 14.05*** No school ICT  − 0.27 3.71  − 5.64** 77.47  − 1.37 18.82  − 7.28** The dependent variable is computed as the average of the ten plausible values in mathematics. Standard errors are clustered at the school level. ***p < 0.01, **p < 0.05, *p < 0.1. All results are weighted. Covariates, indicated in column headers are: Individual factors: gender, age (in months), repeated grade immigrant status, age of arrival; socioeconomic: highest parents’ level of education, highest parents’ level of occupation, books and e-books at home; School includes: types (general, technical, vocational), public or private, and location (city, town or rural) Table 3 evidences that in France, about 56 and 50% of the variations between the base and the full model of the coefficients on No computer and No quiet place, respectively, are due to school characteristics (column 3). Likewise, in Italy, about 50% of the variations in the coefficients on the variables concerning home resources are related to school types. In both countries, socioeconomic factors have a smaller but also important role: they explain about 30% of the variation of coefficients. In Germany, variations in the scores’ gaps related to No computer and No quiet place are especially due to socioeconomic covariates (about 50%, column 2), while variations of No internet coefficient are especially linked to school types and individual characteristics (among which the age of arrival of immigrant students, Table S4). Hence, in countries where tracking starts earlier, France, Italy and Germany, school type variables contribute considerably to explain variations in remote learning gaps. How this happens is shown in detail in the columns concerning General, Technical and Vocational school in Table S4. Interestingly, the negative variation is entirely due to the schools best endowed and with the highest average education levels, which are the general schools, or lyceums. In our data, students who lack the resources needed to learn remotely are also more likely to attend technical and vocational schools, but Table S4 shows that students who lack a computer or a quiet place to study and attend schools where they are common, i.e. lyceums, experience the highest losses in education. This finding is similar to the one seen above regarding cities and rural areas, and is also consistent with the role of technological networks in education. In Spain, individual characteristics (and among them, especially grades repetition, Table S4) explain coefficients’ variations more than in other countries, while socioeconomic factors follow. This indicates that students in Spain who are unable to learn remotely because of a lack of resources are also more likely to repeat grades. These results are consistent with the country’s system of comprehensive schools combined with a high frequency of grades retention. Finally, variations of the four coefficients, from the base to the full model, are small in the United Kingdom, but are mostly due to socioeconomic factors, which contribute to explain about 90% of the total variations. Specifically, Table S4 shows that important among them are the jobs and cultural status of parents (proxied by the number of books at home). In turn, this is consistent with the late tracking and low frequency of grade retention of the country. Interestingly, across countries, being an immigrant student explains more of the variations in the coefficients on No computer or No quiet place than of those on No internet (Table S4). A possible explanation of this discrepancy that immigrant families tend to maintain communication links with people in their home country and, therefore, have a higher access to internet than expected. Another individual variable that leads to somewhat unexpected results is gender. Being female is always negatively correlated with the outcome in mathematics (Table 1), but it counteracts the negative variation of the coefficient on No computer. This can be seen in Table S4, where the coefficient on the female variable is always positive, and is significant in France, Germany and Spain. This is partly due to an average possession of computers for schoolwork that in all countries is higher for females than for males. However, both characteristics, gender and immigrant status explain very small portions of the variation of the coefficients on the variables of interest, smaller than that explained by school types or parents’ occupations. In sum, results up to now show that the gaps in mathematics due to the lack of resources needed to learn remotely are partly explained by covariates, such as school locations, school types, socioeconomic factors or individual characteristics, in proportions that can vary across countries, but also that, after all their effects have been considered, part of the direct correlations between the lack of the resources needed to learn remotely and education outcomes remains strong and significant. Repeating grades and planning to leave education early Not being able to learn remotely can have longer run consequences than those on score gaps. For example, students unable to attend online classes who see their scores falling considerably below those of their peers may form pessimistic expectations regarding the length of their future education. They may plan to drop out of school altogether, or to stop studying when completing their compulsory schooling cycle of secondary school. As already said, we use the question Which of the following do you expect to complete? with which we build a dummy variable taking value one if the student expects to complete at most the lower secondary (ISCED level 2) or upper secondary levels providing direct access to the labour market (ISCED levels 3C or 3B), and zero if the student plans to complete higher levels. Moreover, if falling behind may reduce students’ planned investments in education, repeating grades may reinforce these plans. Hence, we expect students unable to attend remote learning to be more likely to plan to cut their planned investments in education early if they are also likely to repeat grades. We first test separately whether our four variables of interest are associated with the probabilities of forming plans to stop education early and with repeating, and, second, we test whether they are correlated with the joint probabilities of the two events. As stated in Eqs. (2) and (3) above, we use Probit specifications for the first two tests and Bivariate Probit regressions for the latter. In the Probit specification, the coefficients of the marginal probabilities on each variable of interest are in columns 1 to 4 of Table 4. The base regressions include only our four variables of interest, while the full specifications control for all covariates of Eqs. (2) and (3). The results on the Bivariate Probit regressions are in columns 5 and 6. The Rho coefficients report the correlation between the residuals of the regressions having Leaving education early and Repeated grades as dependent variables. Table 4 Repeating grades and planning to leave education early Dependent variable: Probit Bivariate probit Leaving education early Repeated grade Leaving education early &  Repeated grade Base Full Base Full Base Full (1) (2) (3) (4) (5) (6) France No computer 0.06** (0.025) 0.02 (0.021) 0.179*** (0.028) 0.02*** (0.008) 0.04*** (0.009) 0.01 (0.001) No internet 0.01 (0.046) 0.01 (0.50) 0.02 (0.045) 0.00 (0.015) 0.00 (0.008) 0.00 (0.001) No quiet place to study 0.02 (0.021)  − 0.01 (0.02) 0.119*** (0.024) 0.00 (0.008) 0.02*** (0.006) 0.00 (0.001) No school ICT 0.01 (0.013) 0.01 (0.01) 0.04 (0.03) 0.01 (0.011) 0.00 (0.004) 0.00 (0.001) Rho 0.01 (0.04)  − 0.26*** (0.06) Predicted mean y1, y2 0.12 0.11 0.14 0.07 0.10 0.05 Observations 5128 4718 5330 4852 5121 4716 Germany No computer 0.24*** (0.036) 0.09** (0.036) 0.13*** (0.036) 0.04 (0.028) 0.13*** (0.028) 0.04** (0.015) No internet 0.17** (0.083) 0.11** (0.083) 0.07 (0.053) 0.03 (0.060) 0.07* (0.043) 0.03 (0.031) No quiet place to study 0.08** (0.040) 0.08** (0.040) 0.09*** (0.031) 0.04 (0.031) 0.06*** (0.022) 0.01 (0.013) No school ICT 0.01 (0.046)  − 0.01 (0.035) 0.00 (0.029)  − 0.02 (0.022) 0.00 (0.021)  − 0.01 (− 0.001) Rho 0.42*** (0.038) 0.24*** (0.039) Predicted mean y1, y2 0.31 0.19 0.19 0.13 0.10 0.05 Observations 3778 3370 4017 3544 3770 3367 Italy No computer 0.06*** (0.018) 0.03** (0.013) 0.09*** (0.024) 0.04* (0.019) 0.03*** (0.01) 0.01** (0.005) No internet 0.02 (0.023) 0.00 (0.016) 0.04 (0.040) 0.01 (0.032) 0.01 (0.01) 0.00 (0.004) No quiet place to study 0.03** (0.013) 0.00 (0.001) 0.07*** (0.021) 0.03* (0.017) 0.02*** (0.006) 0.00 (0.003) No school ICT 0.01 (0.001) 0.01 (0.001) 0.00 (0.016) 0.02* (0.001) 0.00 (0.004) 0.01* (0.001) Rho 0.49*** (0.049) 0.34*** (0.051) Predicted mean y1, y2 0.07 0.04 0.13 0..09 0.03 0.01 Observations 10,438 9820 10,962 10,256 10,431 9817 Spain No computer 0.15*** (0.013) 0.05*** (0.008) 0.31*** (0.017) 0.186*** (0.018) 0.148*** (0.011) 0.05*** (0.007) No internet 0.04*** (0.013) 0.01 (0.010) 0.15*** (0.03) 0.07** (0.03) 0.047*** (0.011) 0.01* (0.007) No quiet place to study 0.03*** (0.01) 0.015** (0.008) 0.05*** (0.016) 0.01 (0.015) 0.025*** (0.007) 0.01** (0.005) No school ICT 0.01 (0.001) 0.00 (0.005) 0.04*** (0.013) 0.02** (0.011) 0.01** (0.005) 0.00 (0.003) Rho 0.90*** (0.049) 0.76*** (0.029) Predicted mean y1, y2 0.08 0.08 0.25 0.03 0.07 0.07 Observations 33,041 30,801 34,004 31,611 33,030 30,795 United Kingdom No computer 0.14*** (0.02) 0.07*** (0.014) No internet 0.13* (0.073) 0.01 (0.038) No quiet place to study 0.07*** (0.02) 0.04** (0.02) No school ICT 0.03** (0.013) 0.01 (0.009) Predicted mean y1, y2 0.15 0.13 Observations 10,271 8721 The dependent variables, Leaving education early (y1) and Repeated grade (y2)are dichotomous variables taking, respectively, value one when the student plans to leave education early and zero otherwise, and value one when grades are repeated and zero otherwise. Rho coefficients report the correlation between the residuals of the regressions having Leaving education early and Repeated grades as dependent variables. Standard errors are clustered at the school level. ***p < 0.01, **p < 0.05, *p < 0.1. All plausible values employed. All results are weighted and replication weights are taken into account. ‘Base' columns 1, 3 and 5 include only the variables of interest, while 'Full' columns 2, 4 and 6 include all covariates of Eqs. (2) and (3). Margins are computed at mean values of covariates Results from Table 4 show that in all countries the lack of ICT resources, especially No computer, significantly increase the separate probabilities of repeating grades (except for the United Kingdom) and of planning to leave education early. In the full regressions of columns 2 and 4, No computer increases the probability of repeating a grade in France by two percentage points (from the average probability of seven percent, shown by the predicted mean of Repeated grade, in column 4). In Germany it rises the probability of planning to leave education early by nine percentage points (the average being 19%; column 2), while the joint probability of planning to leave education early and repeating a grade rises by four percentage points (on an average of 24%; column 6). In Italy, No computer is associated with a higher probability of planning to leave education early that corresponds to three percentage points and repeating a grade to four, while it increases their joint probability by one percentage point. Similar results on the joint probabilities apply to No school ICT. In Spain most coefficients on our variables of interest are strong and significant. In the fully controlled regressions, No computer is associated with an increase in the joint probability of repeating a grade and planning to leave education early of five percentage points, No internet and No quiet place are related to increases of one percentage point. The Rho coefficient, indicating the degree of correlation between the two probabilities is higher than in the other countries. Finally, in the United Kingdom, No computer and No quiet place to study increase the probability of planning to leave school early by seven and four percentage points, respectively. In sum, being unable to learn remotely is strongly associated with grades repetition, except for the United Kingdom, and with students planning to leave school early. Moreover, it is correlated with the joint probability of both events in the three countries where grades repetition is more frequent, namely Germany, Italy and Spain. Sensitivity and further robustness checks The wide range of controls included in Table 1, the corroborating results of the Gelbach decomposition, Oster tests and subsequent checks on school types and characteristics support the robustness of our findings. However, to further check that the coefficients on our variables of interest are not in fact driven by unobserved family’s socioeconomic factors we interacted our variables of interest with three different levels of these covariates. The underlying hypothesis is that if coefficients on our variables of interest were just absorbing the effect on scores of family conditions, then the coefficients on the variables interacted with a low socioeconomic status level would be negative, indicating that their relationship with scores worsens when socioeconomic conditions are low, while the opposite would hold for higher levels. Table S5 in the Online Appendix reports the coefficients of all our variables of interest interacted with three proxies levels of income: parents’ education, parents’ jobs, and number of books and of e-books at home. The intermediate socioeconomic status is in the intercept. Results show that coefficients on the interacted variables are mostly non-significant, and in the few cases in which they are, their signs do not support the above hypothesis and in several cases go against it.5Among socioeconomic indicators, we also controlled for the number of cars owned by the student’s family, but since this variable entails a substantial loss of observations, we chose not to include it among the covariates of Table S5.6 Another possibility is that lacking the resources needed for learning remotely might matter less when neither the school the student attends possesses ICT tools for education. In this case, the inequality within schools deriving from the scarcity of resources at home should weaken, while that between schools can persist. We tested this hypothesis by interacting the variable No school ICT with the other three variables of interest, but found that the coefficients on these interactions are generally non-significant. As shown in Table S6, one exception is Italy, where in the full regression the coefficient on the interaction of No internet at home and No school ICT is positive and significant (column 5). As expected, not having an internet connection at home is correlated with a smaller gap in education when, everything else given, the school attended does not make use of ICT resources for teaching. Even in the case of Italy, however, when school fixed effects are included into the regression, both coefficients, on No internet and on its interaction with No school ICT shrink and become non-significant (column 6).7 In further checks of the robustness of our findings, we repeated our regressions with supplementary sets of covariates, comprising language spoken at home, different types of ICT resources available at school, and teachers’ digital training. Moreover, we substituted the dependent variable with the scores in reading, and rerun all regressions. Both for mathematics and reading,  repeated all tests with balanced samples based on the observations of the full regressions. Next, we repeated the mafter imputing the values of the missing observations in the full samples. All these checks provided support for our main results and are available upon request. Conclusions This study uses PISA 2018 data on mathematics from five European countries to investigate whether education inequalities can be related to the lack of the resources needed to learn remotely—a computer for schoolwork, an internet connection, a quiet place to study, a school with a platform for online teaching—. After controlling for a wide set of cofactors, fixed effects and different specifications, we find that students lacking these resources score significantly below their peers. Symmetrically, and differently from previous studies pointing at negative or null effects of the use of computers or internet in schooling (Leuven et al. 2007; Malamud and Pop-Eleches 2011; Vigdor et al. 2014; Carter et al. 2017), we find that the relationships between remote learning and education outcomes, both at the students’ and schools’ levels, are positive and significant. Part of the negative gaps are explained by education systems, school types, locations and individual or socioeconomic factors, with the importance of each of them varying across countries. In cities, the use of ICT resources for schooling is more common than in towns and rural areas, and in countries with early tracking, general schools (lyceums) are better endowed, but both register the larger negative gaps of students unable to learn remotely. In countries with late tracking, socioeconomic and individual factors play a more direct role in explaining gap variations, but, also in this case, the correlation is stronger for students with better-off families who lack the resources needed for online learning. Further results from our research show that education inequalities can have long-lasting consequences, especially in countries where grade retention is more frequent. Students not possessing the resources needed for learning remotely and repeating grades are more likely to plan to end their education early or/and drop out from school. These findings point to a more general result: there are technological network externalities in remote and online learning that make the losses of outsiders larger as it spreads and becomes an integral part of education. The positive correlation between the use of these resources and average education levels together with the fact that gaps can remain significant even after all cofactors have been controlled for, imply that even in non-pandemic times governments must decisively promote the use of remote learning at a general level, and at the same time especially focus on those students and schools that lack the resources needed to participate into it. The nature of externalities implies that low-level education equilibria, where both schools and students lack these resources, are self-sustaining. The Covid-19 pandemic accelerated the use of remote learning in several countries but spread unevenly across areas, schools and families, which makes the need of corrective policy actions now even stronger than before. Electronic supplementary material Below is the link to the electronic supplementary material.Supplementary file1 (PDF 336kb) Appendix Table A1 Descriptive statistics France Germany Italy Spain United Kingdom Obs. Mean Std. Dev. Obs. Mean Std. Dev. Obs. Mean Std. Dev. Obs. Mean Std. Dev. Obs. Mean Std. Dev. Math score 6308 495.41 92.57 5451 500.04 95.39 11,785 486.59 93.78 35,943 481.39 88.4 13,818 501.77 93.02 Reading score 6308 492.61 101.18 5451 498.28 105.75 11,785 476.28 96.87 – – – 13,818 503.93 100.21 Leave educ. Early 5930 0.12 0.32 4408 0.31 0.46 10,943 0.06 0.23 34,406 0.09 0.28 12,750 0.13 0.33 Repeated grade 6215 0.17 0.37 4674 0.20 0.4 11,495 0.13 0.34 35,449 0.29 0.45 13,306 0.03 0.16 No computer 6193 0.09 0.29 4711 0.08 0.27 11,485 0.10 0.3 35,391 0.09 0.28 13,250 0.08 0.27 No internet 6203 0.02 0.12 4721 0.02 0.14 11,491 0.03 0.17 35,371 0.02 0.14 13,262 0.01 0.09 No quiet place to study 6186 0.06 0.24 4723 0.05 0.21 11,491 0.09 0.28 35,372 0.07 0.26 13,204 0.11 0.31 No school ICT 5458 0.65 0.48 4718 0.67 0.47 11,291 0.54 0.5 34,738 0.48 0.5 11,331 0.34 0.47 Female 6308 0.49 0.5 5451 0.46 0.5 11,785 0.48 0.5 35,943 0.49 0.5 13,818 0.51 0.5 Age 6308 15.86 0.29 5451 15.83 0.29 11,785 15.77 0.29 35,943 15.84 0.29 13,818 15.76 0.28 Parents' education 6133 4481 11,439 34,925 12,391 Low education level 6133 0.08 0.26 4481 0.21 0.41 11,439 0.15 0.36 34,925 0.16 0.37 12,391 0.03 0.17 Average education level 6133 0.21 0.41 4481 0.25 0.43 11,439 0.42 0.49 34,925 0.14 0.34 12,391 0.32 0.47 High education level 6133 0.71 0.45 4481 0.54 0.5 11,439 0.43 0.5 34,925 0.70 0.46 12,391 0.65 0.48 Parents' occupation level 5806 4437 11,053 34,246 11,992 Low occupation level 5806 0.33 0.47 4437 0.34 0.47 11,053 0.34 0.47 34,246 0.34 0.47 11,992 0.34 0.47 Average occupation level 5806 0.34 0.47 4437 0.33 0.47 11,053 0.33 0.47 34,246 0.33 0.47 11,992 0.33 0.47 High occupation level 5806 0.33 0.47 4437 0.33 0.47 11,053 0.33 0.47 34,246 0.33 0.47 11,992 0.33 0.47 Books at home 6157 4722 11,459 35306 13,196 0–25 books 6157 0.37 0.48 4722 0.25 0.43 11,459 0.27 0.45 35,306 0.25 0.43 13,196 0.36 0.48 26–200 books 6157 0.41 0.49 4722 0.47 0.5 11,459 0.50 0.5 35,306 0.52 0.5 13,196 0.45 0.5 More than 200 books 6157 0.22 0.41 4722 0.28 0.45 11,459 0.23 0.42 35,306 0.23 0.42 13,196 0.19 0.39 e-Books at home 6110 4684 11,409 35,198 13,230 None 6110 0.80 0.4 4684 0.66 0.31 11,409 0.72 0.45 35,198 0.59 0.49 13,230 0.50 0.5 1–2 e-Books 6110 0.17 0.38 4684 0.31 0.46 11,409 0.26 0.44 35,198 0.38 0.48 13,230 0.43 0.5 3 or more e-Books 6110 0.03 0.17 4684 0.03 0.18 11,409 0.02 0.13 35,198 0.03 0.17 13,230 0.07 0.25 Immigrant status 6167 0.14 0.35 4727 0.22 0.42 11,354 0.10 0.3 34,844 0.12 0.33 12,979 0.20 0.4 Age of arrival 6177 0.51 2.29 4798 0.71 2.81 11,479 0.43 1.95 35,419 0.66 2.48 13,293 0.84 2.86 School type 6308 5451 11,785 35,943 13,818 General school 6308 0.64 0.48 5451 0.55 0.5 11,785 0.48 0.5 35,943 0.99 0.1 13,818 1.00 – Technical school 6308 0.30 0.46 5451 0.38 0.49 11,785 0.31 0.46 35,943 – 0.01 13,818 – – Vocational school 6308 0.60 0.24 5451 0.07 0.26 11,785 0.20 0.4 35,943 0.01 0.1 13,818 – – Public school 5602 0.80 0.4 4690 0.96 0.19 11,575 0.96 0.19 34,911 0.68 0.47 11,888 0.34 0.47 Location of school 5602 4663 11,575 34,884 11,859 Location: Rural area 5602 0.03 0.16 4663 0.11 0.11 11,575 0.04 0.19 34,884 0.04 0.21 11,859 0.07 0.26 Location: Town 5602 0.75 0.43 4663 0.72 0.45 11,575 0.72 0.45 34,884 0.59 0.49 11,859 0.62 0.49 Location: City 5602 0.22 0.42 4663 0.27 0.44 11,575 0.24 0.42 34,884 0.36 0.48 11,859 0.31 0.48 Lack of infrastructures 5515 0.29 0.46 4695 0.37 0.48 11,416 0.53 0.5 34,743 0.42 0.49 11,179 0.34 0.47 Inadequate infrastructures 5515 0.28 0.45 4668 0.42 0.49 11,433 0.55 0.5 34,636 0.39 0.49 11,240 0.33 0.47 All plausible values employed. All results are weighted and replication weights are taken into account Table A2 Definition of variables Variable Definition Math score Continuous variable representing the students’ score in mathematics Leaving education early Binary variable taking value 1 when the student plans to complete at most ISCED levels 3C or 3B and 0 otherwise Repeated grade Binary variable taking value 1 when the student has repeated a grade and zero otherwise No computer Binary variable taking value 1 when the student states that she/he does not possess a computer at home to use for schoolwork and 0 otherwise No internet Binary variable taking value 1 when the student states that she/he does not have an access to the internet at home and 0 otherwise No quiet place to study Binary variable taking value 1 when the student states that she/he does not a quiet place to study at home and 0 otherwise No school ICT Binary variable taking value 1 when the school administrator states that the school does not possess an effective online learning support platform and 0 otherwise Female Binary variable taking value 1 for female 0 for male Age Continuous variable representing the students’ age at the time of interview, in years and months Low education level Binary variables representing the highest education level among parents. They take value 1, respectively, if; High, at least one parent has tertiary education, Average, neither parent has more than upper secondary education, Low, both have less than secondary education, and zero otherwise Average education level High education level Low occupational level Binary variables representing the highest occupational status among parents. Based on the ISEI classification of occupations, the occupational level is divided into Low, Average, and High by splitting the observations into tertiles. Dummy variables take value 1 in accordance to each level and 0 otherwise Average occupational level High occupational level 0–25 books Binary variables taking value 1 in correspondence to each number of books at home and 0 otherwise 26–200 books More than 200 books None Binary variables, taking value 1 in correspondence to each number of e-books at home and 0 otherwise 1–2 e-books 3 or more e-books Immigrant status Binary variables taking value 1 if the student is foreign born or his/her citizenship differs from that of the country of the test and 0 otherwise Age of arrival Continuous variable indicating the age of the arrival of the student in years and moths. It equals 0 if the student is native General school Binary variables taking value 1 in correspondence to each type of school and 0 otherwise Technical school Vocational school Public school Binary variables taking value 1 if the student attends a public school, and 0 otherwise Rural area Binary variables indicating taking value 1 in correspondence to each size of the municipality where the school is located and 0 otherwise. City comprises more than 100,000, Town between 100,000 and 3000 Rural area less than 3000 Town City Lack of infrastructure Binary variable taking value 1 if the school administrator states that the physical infrastructure of the school is inadequate and 0 otherwise Inadequate infrastructure Binary variable taking value 1 if the school administrator states that the quality of the physical infrastructure is poor and 0 otherwise 6. 1 However, because of the pre-existing ICT structure provided by the Plan Ceibal, Uruguay could react faster than other countries in the region to school closures during the pandemic; with an action denominated Ceibal at Home, it rapidly implemented remote learning through the country (Ripani 2022). 2 Together with reading, mathematics is one of the two main fields considered in the literature on remote and digital schooling. Since PISA 2018 does not comprise data on reading from Spain, we focused on mathematics. However, we employed the same empirical strategy described below in Sect. 4 with the scores in reading instead of those mathematics, and found very similar results to those of Sect. 5. They are available upon request. 3 While several questions on schools’ material resources are based on the perceptions of school principals, the question we select, on the availability of a platform for online learning, is a factual indicator. 4 Regarding the United Kingdom, the Oster test on No quiet place is δ = 8.6, which indicates a result robust to potential unobserved confounders. 5 Since splitting the variable of interest into three sub-groups lowers the number of observations in each group, with the possible effect of weakening the significance of coefficients on the interacted variables, we repeated these regressions by using the socioeconomic indicators as continuous variables, and reached very similar results. 6 Considering the fully controlled regressions, the percentage of missing observations is approximately 5 for France, 4 for Germany, 1.5 for Italy, 2 for Spain, and 6 for the United Kingdom. As expected, we find that belonging with a family that owns at least one car is positively correlated with school outcomes, but, more interestingly, we also find that, when significant. the correlation is higher when the family owns two cars than when it owns less or more than two. This non-linearity in coefficients could be related to the family’s composition, with two cars suggesting a two-parents’ family, and suggesting also that both parents are income earners. Unfortunately, in PISA 2018 there are no indicators on the composition of the family that could substantiate this hypothesis. Other than its direct correlation with education outcomes, including the number of cars did not significantly alter our main results. 7 We thank an anonymous referee for suggesting the use of the number of cars as a covariate and for indicating the possible combined role of the lack of resources at school and at home. Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References Banerjee AV Shawn C Duflo E Linden L Remedying education: evidence from two randomized experiments in India Q J Econ 2007 122 3 1235 1264 10.1162/qjec.122.3.1235 Beuermann DW Cristia J Cueto S Malamud O Cruz-Aguayo Y One laptop per child at home: short-term impacts from a randomized experiment in Peru Am Econ J Appl Econ 2015 7 2 53 80 10.1257/app.20130267 Cameron AC Trivedi K Microeconometrics: methods and applications 2005 Cambridge Cambridge University Press Carter SP Greenberg K Walker MS The impact of computer usage on academic performance: evidence from a randomized trial at the United States Military Academy Econ Educ Rev 2017 56 118 132 10.1016/j.econedurev.2016.12.005 Comi SL Argentin G Gui M Origo F Pagani L Is it the way they use it? Teachers, ICT and student achievement Econ Educ Rev 2017 56 24 39 10.1016/j.econedurev.2016.11.007 Cristia J Czerwonko A Garofalo P Does technology in schools affect repetition, dropout and enrollment? Evidence from Peru J Appl Econ 2014 17 1 89 111 10.1016/S1514-0326(14)60004-0 Escueta M Nickow AJ Oreopoulo P Quan V upgrading education with technology: insights from experimental research J Econ Lit 2020 58 4 897 996 10.1257/jel.20191507 Fairlie RW The effects of home access to technology on computer skills: evidence from a field experiment Inf Econ Policy 2012 24 3–4 243 253 10.1016/j.infoecopol.2012.06.001 Fairlie RW Bahr PR The effects of computers and acquired skills on earnings, employment and college enrolment: evidence from a field experiment and California UI earnings records Econ Educ Rev 2018 63 51 63 10.1016/j.econedurev.2018.01.004 Fairlie RW Grunberg SH Access to technology and the transfer function of community colleges: evidence from a field experiment Econ Inq 2014 52 3 1040 1059 10.1111/ecin.12086 Fairlie RW London RA The effects of home computers on educational outcomes: evidence from a field experiment with community college students Econ J 2012 122 561 727 753 10.1111/j.1468-0297.2011.02484.x Fairlie RW Robinson J Experimental evidence on the effects of home computers on academic achievement among schoolchildren Am Econ J Appl Econ 2013 5 3 211 240 10.1257/app.5.3.211 Gelbach JB When do covariates matter? And which ones, and how much? J Law Econ 2016 34 2 509 543 Kuhfeld M Soland J Tarasawa B Johnson A Ruzek E Liu J Projecting the potential impacts of COVID-19 school closures on academic achievement Educ Res 2020 49 8 549 565 10.3102/0013189X20965918 Leuven E Lindahl M Oosterbeek H Webbink D The effect of extra funding for disadvantaged pupils on achievement Rev Econ Stat 2007 89 4 721 736 10.1162/rest.89.4.721 Malamud O Cueto S Cristia J Beuermann DW Do children benefit from internet access? Experimental evidence from Peru J Dev Econ 2019 138 41 56 10.1016/j.jdeveco.2018.11.005 Malamud O Pop-Eleches C Home computer use and the development of human capital Q J Econ 2011 126 2 987 1027 10.1093/qje/qjr008 22719135 OECD PISA 2018 Technical report 2018 Paris PISA, OECD Publishing OECD PISA 2018 Results (volume I): what students know and can do 2019 Paris PISA, OECD Publishing Oster E Unobservable selection and coefficient stability: theory and evidence J Bus Econ Stat 2019 37 2 187 204 10.1080/07350015.2016.1227711 Ripani M Vincent-Lancrin S Cobo Romaní C Reimers F Uruguay: Ceibal at Home How learning continued during the COVID-19 pandemic: global lessons from initiatives to support learners and teachers 2022 Paris OECD Publishing Schleicher A (2020) The impact of COVID-19 on education: insights from Education at a Glance 2020. OECD, Paris. https://www.oecd.org/education/the-impact-of-covid-19-on-education-insights-education-at-a-glance-2020.pdf. Accessed 12 Feb 2022 Vigdor JL Ladd HF Martinez E Scaling the digital divide: Home computer technology and student achievement Econ Inq 2014 52 3 1103 1119 10.1111/ecin.12089 Woessmann L International evidence on school tracking: a review Cesifo DICE Rep 2009 7 1 26 34 Yanguas ML Technology and educational choices: evidence from a one-laptop-per-child program Econ Educ Rev 2020 76 1 13 10.1016/j.econedurev.2020.101984

0

PMC9753861

NO-CC CODE

2022-12-16 23:26:19

no

Empirica (Dordr). 2022 Dec 15;:1-30

utf-8

Empirica (Dordr)

10.1007/s10663-022-09556-7

oa_other

==== Front Med Sci Educ Med Sci Educ Medical Science Educator 2156-8650 Springer US New York 1714 10.1007/s40670-022-01714-7 Original Research Impact of the COVID-19 Pandemic on Students’ Motivation in Relation to Asynchronous Anatomy Video Lectures http://orcid.org/0000-0001-7487-3101 Cardoso-Júnior Aloísio 123 http://orcid.org/0000-0001-7740-8408 Faria Rosa Malena Delbone 1 1 grid.8430.f 0000 0001 2181 4888 Postgraduate Program in Pathology, Faculty of Medicine, Federal University of Minas Gerais, Belo Horizonte, MG Brazil 2 grid.441982.2 0000 0004 0643 9452 Medical School, Universidade José Do Rosário Vellano (UNIFENAS), Belo Horizonte, MG Brazil 3 Av. Carandaí 362/1001., Belo Horizonte, MG 30130-060 Brazil 15 12 2022 110 5 12 2022 © The Author(s) under exclusive licence to International Association of Medical Science Educators 2022, Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. Objectives This study aimed to investigate medical students’ motivation in relation to asynchronous anatomy video lectures, carried out during COVID-19 remote teaching. Methods Repeated cross-sectional modified Instructional Materials Motivation Survey questionnaire, validated in Brazil, was applied to 255 students attending the first semester of the undergraduate medical course at the José do Rosário Vellano University, in June 2020 and November 2020. The data were analyzed considering the 95% confidence level as significant (p < 0.05). Results The overall score of motivation attributed by the students was moderate to high (3.7/5, 74%). The same occurred in relation to all dimensions of the instrument: Interest (3.6/5, 72%), Confidence (3.7/5, 74%), Attention (3.5/5, 70%), and Expectation (3.7/5, 74%). Cluster analysis showed that 78% (n = 168) of the students had moderate (72% of the maximum score) or high (86% of the maximum score) degrees of motivation. The influence of social isolation on the students’ emotional state did not affect the overall motivation scores (p = 0.217) or the dimensions of motivation: Interest (p = 0342), Confidence (p = 0.061), Attention (p = 0.625), and Expectation (p = 0.094). Conclusions The students showed high motivation for the asynchronous video lectures of human anatomy. Although the majority of students are highly affected regarding their emotional state, due to the social isolation imposed by the COVID-19 pandemic, this fact did not interfere with the motivation for video lectures, probably due to the high intrinsic motivation that students in the first year have in relation to anatomy. These findings alert to the importance of asynchronous video lectures as an adequate strategy for the teaching and learning of human anatomy. Supplementary Information The online version contains supplementary material available at 10.1007/s40670-022-01714-7. Keywords Motivation Video lectures Anatomy COVID-19 Remote education ==== Body pmcIntroduction The primary objective of education is to provide an environment that promotes an in-depth understanding of the subjects under study, through the individuals’ personal capacities and desires, enhancing their innate skills, as well as those acquired throughout life, resulting in significant learning. Thus, the energy that drives the learning process is motivation [1]. In this environment, motivation is one of the most impactful elements of learning, in a polymorphic context that involves attitudes, strategies, and goals [2]. Among several definitions found in the literature, motivation can be seen as the process in which goal-driven behavior is instigated and sustained [3]. Individuals who are motivated to learn undertake efforts to direct their energy towards attention, concentration, interaction, and development, satisfying their personal motives, values, and expectations [4, 5]. In this sense, Pintrich [6] emphasized the relationship between motivation and cognition in student performance and learning. One approach of achieving an understanding of motivation is the attention, relevance, confidence, satisfaction (ARCS) model proposed by Keller [7]. These four categories promote an overview of the major dimensions of human motivation in the context of learning and how to create strategies to stimulate and sustain motivation [8]. The significant learning of human anatomy, considering the human body structures and their functions, is essential for medical students for the structuring of the knowledge that will be used to understand physiology and pathology, as well as for the development of semiology, diagnostic reasoning, surgical techniques, and the interpretation of imaging tests [9]. In recent years, there has been a growing interest in revising the medical school curricula aiming to improve the teaching of human anatomy. Anatomy is seen as an area of knowledge with a high intrinsic cognitive load due to the high volume of information, which often leads to superficial learning due to the wide scope and little depth attributed to the several learning objectives listed in the traditional curricula [10, 11]. Despite the historical importance of traditional methods based on the dissection of cadavers, its use as an isolated teaching strategy has been supplanted in curricula that favor the integration of basic science learning objectives, such as anatomy, with clinical, surgical, and imaging testing aspects, aiming to contextualize learning and motivate students in the initial stages of the course [9, 12–14]. In this scenario, new teaching–learning tools have been used together with innovative curricular principles, which privilege active learning, the rescue of prior knowledge, the horizontal and vertical integration of knowledge, the spiral allocation of learning objectives, and feedback. Therefore, educational resources such as video lectures, virtual reality, augmented reality, gamification, computerized anatomy tables, and computer-based teaching are being tested and implemented in the teaching of human anatomy [15–20]. Estai and Bunt [9], in a critical review of the literature regarding the several resources and learning strategies used for the teaching of human anatomy, emphasized, among other recommendations, that there is no single model capable of meeting all the curricular needs. Therefore, they pointed out that the best way to teach anatomy is through the combination of several mutually complementary instructional tools. This versatility, regarding the combination of teaching–learning resources, can allow the adoption of the most convenient methods for each curricular context, considering the peculiarities of each course, as well as both the economic and social aspects. This variety of teaching–learning resources was crucial for the continuity of human anatomy teaching during the pandemic caused by the new coronavirus (COVID-19), which broke out in the beginning of 2020 [21]. Considering the imminent need for social isolation, the immediate adaptation of medical courses became imperative and, consequently, also of the teaching–learning strategies of human anatomy, on an emergency [22]. Among the several resources that can be integrated into the remote teaching of human anatomy, many medical courses have chosen to use narrated video lectures, included in virtual learning environments, for their asynchronous use by the students [23–27]. The autonomy provided by these video lectures can play an important role in increasing student’s motivation as demonstrated by self-determination theory [28, 29]. In turn, the students’ perception regarding the teaching of anatomy, in these pandemic times, is a beneficial one and supports the justification for this research [30, 31]. The literature lacks studies that specifically evaluate students’ motivation for asynchronous human anatomy video lectures. Therefore, considering its importance for the teaching of human anatomy during the COVID-19 pandemic, this study aimed to assess (1) The motivation of students attending the Medical School at José do Rosário Vellano University (UNIFENAS-BH) for this instructional activity, (2) the influence of sociodemographic factors on students’ motivation, and (3) the influence of the students’ self-declared emotional state during social isolation due to the COVID-19 pandemic, on their motivation for asynchronous video lectures of human anatomy. The results of the study may pave the way for greater use of this educational strategy and support new studies on the topic. Methods Subject Recruitment The target population (n = 255) consisted of all students attending the first semester of the undergraduate medical course at José do Rosário Vellano University (UNIFENAS-BH), Belo Horizonte Campus, during the academic year of 2020. The students were invited to fill out the questionnaire, aimed at measuring their motivation regarding the asynchronous video lectures of human anatomy taught during the social isolation period, as well as evaluating the perception of their emotional state at that time. The inclusion criteria comprised being regularly enrolled at the medical school and their wish to participate in the study. The exclusion criteria comprised the wish to leave the study, the lack of adherence to the protocol, and being of foreign origin, which could hinder the interpretation of the questionnaire items. Curricular Context The undergraduate medical course at UNIFENAS-BH, founded in 2003, adopts the problem-based learning methodology. Human anatomy is studied over four semester terms, during the first 2 years of medical school. Synthetic models, a digital anatomy table, imaging tests, videos of surgical procedures, and prosection are used in the teaching of anatomy. The curriculum consists of thematic blocks that carry out the horizontal and vertical integration of knowledge. The undergraduate medical course duration consists of 12 semesters, divided into pre-clinical, clinical, and internship stages. Asynchronous Video Lectures During the social isolation period, which started in mid-March 2020, the medical course at UNIFENAS-BH chose to use asynchronous video lectures to maintain the human anatomy educational strategy. The teachers of this teaching strategy recorded and made lectures available using the following systems:Recording of video lectures in the Open Broadcaster Software (OBS)®, version 25.0.4 Use of laboratory practice scripts in human anatomy, contained in the student guide, for the standardization of video lectures, as follows: Introduction Learning objectives Checklist of structures for identification Oral presentation, illustrated by PowerPoint slides, with anatomical atlas pictures, focusing on theoretical and practical aspects in an integrated manner Closure of the activity with discussion of a clinical, surgical, or imaging case Recommended bibliography Uploading the video lectures to the institutional Google Drive platform and providing the access link, weekly, in the virtual learning environment (Moodle Unifenas), according to the thematic blocks programming After being uploaded to the institutional Google Drive platform, the video lectures were available to the students for consultation on demand throughout the ongoing thematic block. There was no access control or any type of mechanism that forced students to watch the video lectures. However, the discussions held at the synchronous meetings gave teachers the perception of their use by the students. Moreover, watching the video lectures was important for the learning of the proposed objectives and achieving proficiency in summative exams. The video lectures were not modified between the first and second semesters of 2020. All students used the same form of access and were able to watch the same recordings. First semester of 2020 students participated in the face-to-face course for 4 weeks, prior to the start of social isolation. In the second semester 2020, until the data collection was carried out for this study, the entire course took place remotely. To continue the anatomy course during social isolation, in addition to these video lectures, students participated in synchronous meetings, carried out weekly through the Google Meet application, with the teachers of human anatomy. These meetings were aimed at solving doubts and discussing clinical cases and imaging tests related to that week’s learning objectives. Data Collection The Instructional Materials Motivation Survey (IMMS) questionnaire, based in the ARCS model of motivational design, was developed to measure students’ motivation regarding instructional materials that have self-directed characteristics, and it has been used in the assessment of video lectures [32, 33]. Aiming to assess the students’ motivation in this study, the translation and transcultural adaptation of the IMMS to Brazilian Portuguese was initially carried out [34]. Subsequently, its psychometric validation generated the IMMS version validated in Brazil (IMMS-BRV), consisting of 25 items divided into four dimensions: Interest, Confidence, Attention, and Expectation (Online Resource 1). The internal consistency reliability, measured by Cronbach’s alpha coefficient, for the complete IMMS-BRV instrument, is 0.95, being 0.93 for the Interest dimension, 0.87 for the Confidence dimension, 0.76 for the Attention dimension and 0.78 for the Expectation dimension [35, 36]. Its purpose is to measure the respondent’s motivation using the Likert scale, with the following score: (1) I totally disagree; (2) I partially disagree; (3) I neither disagree nor agree; (4) I partially agree; and (5) I totally agree. In addition to the IMMS-BRV, students answered a complementary questionnaire with two items about their self-perceived motivation for human anatomy video lectures and one item about the influence of social isolation measures on their emotional state. The items were as follows: (1) I feel naturally motivated for video lectures in human anatomy, and (2) video lectures in human anatomy make me motivated (1, I strongly disagree; 2, I partially disagree; 3, I neither disagree nor agree; 4, I partially agree; 5, I totally agree); (3) did the social isolation measures affect your emotional state? (1, They did not affect; 2, They affected a little; 3, They affected moderately; 4, They affected a lot; 5, They totally affected). Data collection was performed once in the June 2020 and once in November 2020, using the Google Forms, at equivalent stages of the course, so that there was enough time for students to be exposed to human anatomy video lectures prior to their motivation assessment. The students were invited to participate by email and WhattsApp. Ethical Approval The present study was approved by the research ethics committee of José do Rosário Vellano University (UNIFENAS-BH) (Approval ID: 03,461,718.0.0000.5143). The informed consent form was applied to all research subjects. Data Analysis To describe the results of the assessed variables, the 25th and 75th percentiles, the median, mean, standard deviation (SD), and percentage values were presented. The relationship between two categorical variables was assessed using the chi-square test. Fisher’s exact test was used to evaluate the association and comparison of the groups regarding the proportion of occurrence of a certain event of interest (categorical type variable) for small samples. The comparison between two independent groups, in relation to a quantitative variable, was performed using the non-parametric Mann–Whitney test. Under the same condition, the comparison between three independent groups was performed using the Kruskal–Wallis non-parametric test (quantitative variable). The cluster analysis, based on the K-means method, was used to determine different profiles of medical students regarding the dimensions of the IMMS-BRV that assess motivation. Statistical significance was set at p < 0.05. Statistical analyses were performed using the SPSS statistical package software, version 20.0 (IBM Corp, Armonk, NY). Results Using a convenience sampling method, from the target population (n = 255), 98 students were recruited in June 2020, and 117 students were recruited in November 2020, totaling 215 (84.3%) undergraduate first semester medical students, of which 147 (68.4%) were women and 68 (31.6%) were men. Regarding age, 47.4% were aged between 17 and 19 years, 25.1% between 20 and 22 years, 13% between 23 and 25 years, 7.5% between 26 and 30 years, and 7% were at least 31 years old. Most students (83.7%) did not have another higher education degree. Of the ones who had a prior higher education degree, the academic courses comprised the following areas of knowledge: biological sciences/health area in 48.6%; social/human sciences in 28.6%; and exact sciences in 22.8%. The students’ motivation scores, measured by the IMMS-BRV, are summarized in Table 1. It can be observed that the overall score of motivation attributed by the students was moderate to high (3.7/5, 74%), demonstrating the students’ good motivation regarding the assessed video lectures. The same occurred in relation to all dimensions of the instrument: Interest (3.6/5, 72%), Confidence (3.7/5, 74%), Attention (3.5/5, 70%), and Expectation (3.7/5, 74%).Table 1 Student motivation in relation to video lectures measured by the Instructional Materials Motivation Survey validated in Brazil (IMMS-BRV) Motivation dimensions Descriptive measures Minimum–maximum Median (P25–P75) Mean ± SD Overall 1.3–5.0 3.8 (3.3–4.3) 3.7 ± 0.8 Interest 1.2–5.0 3.7 (3.0–4.2) 3.6 ± 0.8 Confidence 1.3–5.0 3.8 (3.0–4.5) 3.7 ± 0.9 Attention 1.0–5.0 3.7 (2.7–4.3) 3.5 ± 1.0 Expectation 1.2–5.0 3.7 (3.3–4.2) 3.7 ± 0.7 Total n = 215 students The influence of sociodemographic data on the motivation and overall motivation dimensions, measured by the IMMS-BRV, can be seen in Table 2. The statistical analysis showed significantly higher scores regarding the Attention dimension in the female group, in comparison to the male group (p = 0.023). Similarly, the group that had a prior higher education degree showed higher scores in the Attention dimension (p = 0.006) in relation to the group that did not have it. However, there was no difference in the comparison between the area of knowledge of the prior degree and motivation.Table 2 Influence of sociodemographic data on motivation measured by the Instructional Materials Motivation Survey validated in Brazil (IMMS-BRV) Variables Motivation factors Overall motivation Interest Confidence Attention Expectation Age group   17 to 19 years 3.7 ± 0.8 3.9 (3.2; 4.3) 3.5 ± 0.8 3.7 (2.8; 4.2) 3.6 ± 0.9 3.8 (3.0; 4.3) 3.4 ± 1.0 3.7 (2.7; 4.3) 3.7 ± 0.7 3.8 (3.1; 4.2)   20 to 22 years 3.7 ± 0.7 3.7 (3.2; 4.3) 3.6 ± 0.8 3.5 (2.8; 4.2) 3.6 ± 1.0 3.6 (2.8; 4.4) 3.5 ± 1.0 3.3 (2.7; 4.3) 3.7 ± 0.6 3.6 (3.2; 4.2)   23 to 25 years 3.9 ± 0.7 3.9 (3.4; 4.5) 3.8 ± 0.9 3.8 (3.0; 4.5) 3.9 ± 0.9 4.0 (3.3; 4.8) 3.5 ± 1.0 3.7 (2.7; 4.3) 3.8 ± 0.6 3.8 (3.4; 4.4)   26 years or older 3.8 ± 0.8 3.8 (3.3; 4.4) 3.7 ± 0.8 3.7 (3.0; 4.3) 3.9 ± 0.8 4.0 (3.3; 4.5) 3.9 ± 0.8 3.7 (3.3; 4.7) 3.8 ± 0.6 3.8 (3.4; 4.3) p 0.570* 0.610* 0.179* 0.158* 0.511* Gender   Female 3.8 ± 0.7 3.8 (3.3; 4.3) 3.6 ± 0.8 3.7 (3.0; 4.2) 3.8 ± 0.9 3.8 (3.3; 4.5) 3.5 ± 1.0 3.7 (2.7; 4.3) 3.8 ± 0.6 3.8 (3.3; 4.2)   Male 3.6 ± 0.8 3.8 (3.1; 4.3) 3.5 ± 0.9 3.6 (2.8; 4.3) 3.5 ± 0.9 3.5 (2.8; 4.3) 3.5 ± 0.9 3.7 (2.8; 4.3) 3.6 ± 0.7 3.7 (3.1; 4.1) p 0.143** 0.273** 0.023** 0.852** 0.095** Prior higher education degree   No 3.7 ± 0.8 3.8 (3.2; 4.3) 3.6 ± 0.8 3.7 (3.0; 4.2) 3.6 ± 0.9 3.8 (3.0; 4.3) 3.5 ± 1.0 3.7 (2.7; 4.3) 3.7 ± 0.7 3.7 (3.2; 4.2)   Yes 3.7 ± 0.7 3.8 (3.3; 4.2) 3.6 ± 0.8 3.7 (3.0; 4.3) 4.0 ± 0.8 4.3 (3.5; 4.8) 3.7 ± 0.9 3.7 (3.0; 4.7) 3.8 ± 0.6 3.8 (3.4; 4.2) p 0.764** 0.993** 0.006** 0.284** 0.574** Area of knowledge of the prior higher education degree   Exact sciences 3.6 ± 0.5 3.8 (3.1; 4.2) 3.7 ± 0.9 3.9 (3.1; 4.5) 4.2 ± 0.6 4.3 (3.9; 4.7) 3.9 ± 1.1 4.2 (3.4; 4.7) 3.9 ± 0.5 3.8 (3.5; 4.3)   Biological sciences/health area 3.5 ± 0.9 3.4 (3.1; 4.0) 3.5 ± 0.9 3.3 (3.0; 4.3) 4.0 ± 0.9 4.0 (3.4; 4.8) 3.8 ± 0.9 3.7 (3.0; 4.7) 3.6 ± 0.7 3.8 (3.2; 3.9)   Social sciences/humanities 4.1 ± 0.5 4.0 (3.8; 4.5) 3.7 ± 0.7 3.8 (30; 4.4) 4.0 ± 0.7 4.1 (3.5; 4.8) 3.2 ± 0.8 3.7 (2.6; 3.8) 3.9 ± 0.4 3.8 (3.6; 4.2) p 0.085* 0.616* 0.829* 0.195* 0.483* Total n = 215; the presented values refer to the mean ± standard deviation and the percentiles P50 (P25; P75); the significance probabilities (p) refer to the Kruskal–Wallis test (*) and the Mann–Whitney test (**) The bolded numbers reached statistical significance The influence of self-reported motivation by the students and their emotional state on motivation scores, measured by the IMMS-BRV, is shown in Table 3. The group that did not feel naturally motivated for human anatomy video lectures was the one with significantly lower scores in all dimensions (p < 0.001) and overall motivation (p < 0.001), and the group that felt naturally motivated for the video lectures in human anatomy was the one with significantly higher scores (p < 0.001). A similar result was observed in relation to the fact that video lectures generate motivation in the students. The group that agreed that video lectures generated motivation had the highest values in all dimensions (p < 0.001) and overall motivation (p < 0.001), and the group that disagreed that video lectures generated motivation had the lowest values (p < 0.001). In turn, the influence of social isolation measures on the emotional state, self-declared by the students, did not interfere with the dimensions of motivation and overall motivation for the asynchronous anatomy video lectures.Table 3 Influence of self-declared motivation and emotional state due to social isolation on motivation measured by the Instructional Materials Motivation Survey validated in Brazil (IMMS-BRV) Variables Motivation factors Overall motivation Interest Confidence Attention Expectation I feel naturally motivated for video lectures on human anatomy   I disagree (1 e 2) 3.0 ± 0.7 3.0 (2.6; 3.4) 2.8 ± 0.7 2.8 (2.3; 3.3) 3.1 ± 0.9 3.3 (2.3; 3.8) 2.8 ± 0.8 2.7 (2;3.3) 3.0 ± 0.6 3.1 (2.7; 3.4)   I neither agree nor disagree (3) 3.8 ± 0.5 3.9 (3.6; 4.3) 3.7 ± 0.6 3.7 (3.3; 4) 3.8 ± 0.8 3.8 (3.3; 4.5) 3.7 ± 0.9 3.7 (3;4.3) 3.8 ± 0.5 3.8 (3.5; 4.1)   I agree (4 e 5) 4.1 ± 0.5 4.3 (3.8; 4.6) 4.0 ± 0.7 4.2 (3.5; 4.5) 4.0 ± 0.8 4.0 (3.3;4.5) 3.9 ± 0.9 4.0 (3.3;4.7) 4.1 ± 0.5 4.1 (3.7; 4.4) p  < 0.001  < 0.001  < 0.001  < 0.001  < 0.001 Human anatomy video lectures generate motivation in me   I disagree (1 e 2) 2.9 ± 0.7 2.9 (2.5;3.3) 2.9 ± 0.8 2.8 (2.3;3.3) 3.1 ± 0.9 3.0 (2.3;3.8) 2.8 ± 0.9 2.7 (2.0;3.5) 3.0 ± 0.6 3.1 (2.7;3.3)   I neither agree nor disagree (3) 3.8 ± 0.5 3.8 (3.5;4.2) 3.6 ± 0.7 3.7 (3.1;4) 3.6 ± 0.8 3.5 (3.2;4.1) 3.5 ± 0.9 3.3 (2.7;4.3) 3.7 ± 0.5 3.7 (3.4;4.1)   I agree (4 e 5) 4.2 ± 0.5 4.3 (3.8;4.6) 4.0 ± 0.6 4.2 (3.7;4.5) 4.1 ± 0.7 4.3 (3.5;4.5) 3.9 ± 0.8 4.0 (3.3;4.7) 4.1 ± 0.4 4.2 (3.8;4.4) p  < 0.001  < 0.001  < 0.001  < 0.001  < 0.001 Did the social isolation measures affect your emotional state?   They did not affect (1) 4.1 ± 0.6 3.9 (3.6;4.6) 4.1 ± 0.6 4.0 (3.6;4.6) 3.9 ± 1.1 3.9 (3.4;4.8) 3.9 ± 0.7 3.9 (3.3;4.7) 4.0 ± 0.6 3.8 (3.6;4.7)   They affected a little (2 and 3) 3.8 ± 0.8 3.9 (3.1;4.3) 3.7 ± 0.8 3.8 (3.2;4.3) 3.7 ± 0.8 3.8 (3.0;4.5) 3.7 ± 0.8 3.7 (3.2;4.3) 3.8 ± 0.6 3.9 (3.3;4.2)   They affected a lot (4 and 5) 3.7 ± 0.8 3.8 (3.3;4.3) 3.5 ± 0.8 3.7 (2.8;4.2) 3.6 ± 0.9 3.8 (3.0;4.5) 3.4 ± 1.0 3.7 (2.7;4.3) 3.7 ± 0.7 3.7 (3.2;4.2) p 0.342 0.061 0.625 0.094 0.217 Total n = 215 students The values shown here refer to the mean ± standard deviation and the percentiles P50 (P25; P75) The significance probabilities (p) refer to the Kruskal–Wallis test The bolded numbers reached statistical significance The cluster analysis, based on the four dimensions of the IMMS-BRV, was carried out to identify different student profiles in relation to the motivation dimensions, with three distinct clusters being identified: cluster 1 (21.9% of students) with low motivation scores (grade I); cluster 2 (41.4% of students) with moderate motivation scores (grade II); and cluster 3 (36.7% of students) with high motivation scores (grade III). The mean overall motivation scores were 2.8 in grade I; 3.6 in grade II, and 4.3 in grade III (Table 4).Table 4 Mean scores of the Instructional Materials Motivation Survey validated in Brazil (IMMS-BRV) in relation to the three student clusters Factors Degree of motivation Analysis of variance Grade I Grade II Grade III p Conclusion Interest 2.8 3.7 4.4  < 0.001 I < II < III Confidence 2.5 3.6 4.2  < 0.001 I < II < III Attention 2.8 3.5 4.4  < 0.001 I < II < III Expectation 2.3 3.3 4.4  < 0.001 I < II < III Overall motivation 2.8 3.6 4.3  < 0.001 I < II < III n (%) 47(21.9) 89 (41.4) 79 (36.7) Total n = 215 students n number of students in the cluster Cluster analysis p ➔ probability of significance of the Kruskal–Wallis test The bolded numbers reached statistical significance The association between sociodemographic data and motivation clusters is shown in Table 5. As observed in the table, there was a significant association between gender and having a previous higher education degree. The female group showed a lower percentage of students with low motivation (grade I) in comparison with the male group (p = 0.040). Similarly, a lower percentage of students with grade I motivation was identified in the group with a prior higher education degree (p = 0.040). These findings corroborate the ones shown in Table 2.Table 5 Association between sociodemographic data and motivation clusters Variables Grade p I II III Age group   17 to 19 years 25 (24.5%) 41 (40.2%) 36 (35.3%) 0.361*   20 to 22 years 15 (27.8%) 22 (40.7%) 17 (31.5%)   23 to 25 years 5 (17.8%) 12 (42.9%) 11 (39.3%)   26 years or older 2 (6.4%) 14 (45.2%) 15 (48.4%) Gender   Female 25 (17.0%) 65 (44.2%) 57 (38.8%) 0.040*   Male 22 (32.4%) 24 (35.2%) 22 (32.4%) Prior higher education degree   No 45 (25.0%) 71 (39.4%) 64 (35.6%) 0.040*   Yes 2 (5.7%) 18 (51.4%) 15 (42.9%) Area of knowledge of the prior higher education degree   Exact sciences 1 (12.5%) 3 (37.5%) 4 (50.0%) 0.781**   Biological sciences/health area 1 (5.9%) 9 (52.9%) 7 (41.2%)   Social sciences/humanities 0 (0.0%) 6 (60.0%) 4 (40.0%) Database: 215 students The p value refers to the probability of significance of the chi-square test (*) or Fisher’s exact test (**) The bolded numbers reached statistical significance Discussion This study aimed to provide an overview of the students’ motivation regarding an instructional tool that was suddenly implemented, in an atypical period and full of uncertainties, having as background a teaching strategy that was very appreciated in the first period of the medical course: human anatomy [37]. This scenario certainly comprised a genuine “trial by fire” for the asynchronous video lectures taught during the assessed period. Although there are studies showing the good acceptance of this tool during the social isolation period, as reported by Srinivasan [31], the students also expressed their concern about the lack of face-to-face practical lectures and their frustration at the loss of the opportunity for dissection and prosection [24, 30]. The study by Sbayeh et al. [37] showed the great importance given by medical students, professors of anatomy, and clinicians of two Irish courses regarding practical lectures and the consequences of the study of anatomy in relation to issues such as professionalism, teamwork, and ethics. Similarly, a study carried out to assess the motivation of students attending the medical course at Queen’s University (Belfast, UK) regarding dissection, using the IMMS as a measuring instrument, found a mean overall score of 4.21 (± 0.99), corroborating the great importance of this strategy for the students [38]. On the other hand, remote teaching has also represented an environment of intense development and training of remote teaching–learning skills. Therefore, the assessment of the educational practices implemented during the COVID-19 pandemic must be carried out to correct directions and identify new opportunities that might be useful beyond the social isolation phase, aiming at a possible appreciation of hybrid teaching. A recent study showed that 87.5% of anatomy students at the University of Singapore School of Medicine, assessed through a questionnaire designed to assess their satisfaction with learning through the Zoom platform, were satisfied with remote teaching [31]. The results of the present study, regarding the students’ overall motivation, showed that the mean score measured by the IMMS-BRV was 3.7, which corresponds to 74% of the total possible score. Hence, it can be considered that students expressed a good motivation mean in relation to asynchronous video lectures in the current social context, as also demonstrated in other studies [39]. Likewise, the mean scores in the four dimensions of motivation homogeneously varied between 3.5 (Attention) and 3.7 (Confidence and Expectation). These findings agree with the results of the study by Huang and Hew [33], which measured, using IMMS, the motivation of students participating in massive online courses (Coursera, Open2study, and Khan Academy), which also used asynchronous video lectures. In their study, the overall motivation score was 3.6, and in the Attention and Confidence dimensions, it was 3.5 and 3.7, respectively. This good motivational behavior is even more evident in the analysis of clusters, where 78.1% of students showed moderate (grade II) or high (grade III) degrees of motivation. A total of 41.4% of students were in the cluster comprising students with grade II motivation (moderate), whose mean overall motivation score was 3.6 (72% of the maximum score), whereas 36.7% of students were in the cluster comprising students with grade III motivation (high), whose mean overall motivation score was 4.3 (86% of the maximum score). Only 2.9% of students were in the cluster with grade I motivation, in which the mean overall score was 2.8 (56% of the maximum score). In this sense, Singh and Ming [40] carried out a study to investigate the attitudes and learning of first-year medical students regarding video lectures used to teach macroscopic anatomy. One cohort of students was submitted to face-to-face classes and the other to asynchronous video lectures, which had the same content. The video lectures group performed significantly better in the tests and evaluated the tool positively. The adequate motivation observed for the anatomy video lectures can also be analyzed in the light of the characteristics of the current generation of students. Barry et al. [41], in the article Anatomy education for the YouTube generation, observed that 78% of the included medical students used YouTube video lectures as a study source, with 50% of the students using them weekly. Similarly, 78% of students reported that the videos were useful, very useful, or extremely useful in helping them understand anatomy. In the present study, all subjects are from Gen Y and Gen Z. This fact may explain the good motivation founded in the target population and the non-difference in the comparison between the age and motivation. When the motivation was analyzed from the perspective of sociodemographic data, regarding the Attention dimension, the scores were significantly higher in the female gender (p = 0.023) and in the group that had a prior higher education degree (p = 0.006). These results were corroborated by the cluster analysis, which showed a lower percentage of students with low motivation (grade I) in the female group than in the male group (p = 0.040). Similarly, a lower percentage of students with grade I motivation was identified in the group with a prior higher education degree (p = 0.040). It is noteworthy that the Attention dimension has items that directly assess the instructional activity (asynchronous video lectures) and its strategies to obtain and maintain attention based on curiosity, with this dimension being considered a personal characteristic and a prerequisite for learning. Therefore, it is necessary that the interface of the video lectures stimulate and sustain the students’ attention [42, 43]. Indeed, the role of the individual’s gender in shaping motivation for learning has been widely discussed in the literature. In a fruitful review article on gender differences regarding motivation, focusing on the expectancy-value theories, causal attribution, self-efficacy, and achievement goals, Meece et al. [44] concluded that the differences are domain-specific, which allows us to highlight the significantly positive result found in the present study, only in the Attention dimension of motivation, in relation to the female gender. Several empirical studies on motivation and learning, in foreign language online courses, have also observed greater motivation in female groups [45]. Nonetheless, in a study involving the assessment of motivation in an online learning environment, it was also verified that motivation was significantly higher in females, assessed by the IMMS dimensions, including Attention [46]. In this study, the author lists several works supporting the fact that women are more engaged and have better learning results than men in the online learning environment. Similarly, a study assessed the behavior of 2927 students regarding 18,144 video lectures related to 13 online modules, from Yasar University, in Turkey. In this study, a significant difference was observed (p < 0.001), in favor of women, in relation to the variable that evaluated whether students fully watched the videos [47]. This finding is completely in agreement with the IMMS-BRV Attention dimension, which verifies whether the instructional material, in the case of this research the asynchronous video lectures of human anatomy, can arouse the students’ curiosity and sustaining their attention throughout the activity. The higher score in the Attention dimension, found in the group of students with a prior higher education degree, may reflect their deeper perception regarding the value of anatomy for their careers, due to their greater maturity and the practical experiences they had in their previous graduation. Aiming to verify the association of the students’ perception of their motivation regarding asynchronous human anatomy video lectures and the motivation measured by a validated instrument (IMMS-BRV), two items were included in the complementary questionnaire, both to be answered using the Likert scale: the first one had a phraseology aimed at detecting intrinsic aspects of motivation (I feel naturally motivated for video lectures on human anatomy). The second item aimed to capture extrinsic aspects of motivation (human anatomy video lectures generate motivation in me). Although they are simple items, which were not submitted to the evaluation of their psychometric properties, the students’ responses showed total agreement with the scores measured by IMMS-BRV. These findings can be seen, a priori, as confirmation of the validity of IMMS-BRV for the measurement of motivation in the studied population. In turn, the influence of social isolation on the emotional state did not affect the overall motivation scores (p = 0.217) and the dimensions of the IMMS-BRV: Interest (p = 0.342); Confidence (p = 0.061); Attention (p = 0.625), and Expectation (p = 0.094). These findings are in agreement with studies of which results showed that although the medical students demonstrated a high degree of anxiety during remote emergency teaching, this behavior, related to the emotional state, is not indiscriminately perceived in relation to all the strategies used in online study. In this sense, asynchronous video lectures and synchronous meetings are being well rated and have not been affected by the emotional state [25, 48, 49]. Still in this regard, a study that included medical and dentistry students attending the first year of the course showed that although 69% of the students felt a lack of “self-motivation,” the anatomy video lectures were appreciated by most students, who felt free to learn at their own pace, taking advantage of the possibility to pause, go back and watch again, as many times as necessary. In this study, only 9% of the participants said they were dissatisfied with the anatomy video lectures. In conclusion, the authors emphasized the gain that remote education is generating in terms of self-directed learning, which can also be highlighted in relation to self-determination [50]. It is interesting to note that, despite the “damage” caused by the impediments to practical face-to-face study, the emergency model adopted by UNIFENAS-BH also had positive and motivating aspects. Therefore, one can affirm that the asynchronous video lectures, available for study prior to synchronous meetings, mimicked the flipped classroom methodology, widely used and well rated in online courses, as it is able to promote aspects of the students’ intrinsic motivation and self-determination [40, 51–53]. This online hybrid anatomy curriculum was viewed as the most effective method because it incorporates the best features of synchronous and asynchronous components [54]. Indeed, intrinsic motivation is related to positive learning environment perceptions and is positively correlated with perceived academic rank [55]. Limitations of the Study The main limitation of this study is related to the studied population. For the motivation assessment to be carried out during the two semesters of 2020, without including the same students again, the study considered only the students attending the first semester of the medical course. In addition, the evaluation of the emotional state through a single question, instead of a validated measurement instrument, also constituted a limitation of the study. However, this strategy was chosen because the primary objective was to assess motivation through the application of IMMS-BRV, and the incorporation of another questionnaire with several items could cause exhaustion in the students and impair the quality of responses. Conclusions This study evaluated an important factor related to meaningful learning of human anatomy. The main conclusions from the study are:The students showed high motivation for the asynchronous video lectures of human anatomy. Although the majority of the participants were very affected by their emotional state due to the social isolation imposed by the COVID-19 pandemic, this fact did not interfere with the motivation for the video lectures, probably due to the intrinsic motivation that students attending the first year have regarding the study of anatomy, as they value its importance for good medical training and due to their curiosity regarding the in-depth knowledge of the human body. The autonomy given to students to watch the asynchronous video lectures, managing the pace of their own study, revising it, when necessary, is probably associated with their motivation for the asynchronous video lectures. These findings alert to the importance of asynchronous video lectures as an adequate strategy for the teaching and learning of human anatomy, suggesting the need for further studies to develop motivating formats adapted to each educational context. Supplementary Information Below is the link to the electronic supplementary material.Supplementary file1 (PDF 72 KB) Acknowledgements We would like to thank the medical students who participated in this research. Author Contribution Aloísio Cardoso-Júnior, conceptualization, methodology, investigation, data curation, and writing—original draft preparation. Rosa Malena Delbone Faria, conceptualization, methodology, and supervision. Declarations Ethical Approval The research ethics committee of José do Rosário Vellano University (UNIFENAS-BH) approved this study (Approval ID: 03461718.0.0000.5143). Consent to Participate Implied consent was obtained from all individual participants included in the study. Conflict of Interest The authors declare no competing interests. Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References 1. Schunk DH Introduction to the special section on motivation and efficacy J Educ Psychol 1990 82 3 6 10.1037/h0092681 2. Donald JG Motivation for higher-order learning New Dir Teach Learn 1999 1999 27 35 10.1002/tl.7803 3. Schunk D Self-efficacy and academic motivation. Educational Psychologist Informa UK Limited 1991 26 207 231 4. Wigfield A Expectancy-value theory of achievement motivation: a developmental perspective Educ Psychol Rev 1994 6 49 78 10.1007/BF02209024 5. Körhasan ND The place of learning quantum theory in physics teacher education: motivational elements arising from the context Kuram ve Uygulamada Egitim Bilimleri 2015 15 1087 1101 6. Pintrich PR A conceptual framework for assessing motivation and self-regulated learning in college students Educ Psychol Rev 2004 16 385 407 10.1007/s10648-004-0006-x 7. Keller JM Development and use of the ARCS model of instructional design J Instr Dev 1987 10 2 10 10.1007/BF02905780 8. Keller JM. Motivational design for learning and performance: the ARCS model approach. Keller JM, editor. New York: Springer; 2010. 9. Estai M Bunt S Best teaching practices in anatomy education: a critical review Ann Anat 2016 208 151 157 10.1016/j.aanat.2016.02.010 26996541 10. Qiao YQ Shen J Liang X Ding S Chen FY Shao L Using cognitive theory to facilitate medical education BMC Med Educ 2014 14 1 7 10.1186/1472-6920-14-79 24387322 11. Taylor DCM, Hamdy H. Adult learning theories: implications for learning and teaching in medical education: AMEE Guide No. 83. Medical Teacher. 2013;35:e1561–72. 12. Ghosh SK Cadaveric dissection as an educational tool for anatomical sciences in the 21st century Anat Sci Educ 2017 10 286 299 10.1002/ase.1649 27574911 13. Ghosh SK Human cadaveric dissection: a historical account from ancient Greece to the modern era Anat Cell Biol 2015 48 153 169 10.5115/acb.2015.48.3.153 26417475 14. Williams SR Thompson KL Notebaert AJ Sinning AR Prosection or dissection: which is best for teaching the anatomy of the hand and foot? Anat Sci Educ 2019 12 173 180 10.1002/ase.1808 30118189 15. Gentry SV Gauthier A Ehrstrom BLE Wortley D Lilienthal A Car LT Serious gaming and gamification education in health professions: systematic review J Med Internet Res 2019 21 1 20 10.2196/12994 16. Ang ET Chan JM Gopal V Li SN Gamifying anatomy education Clin Anat 2018 31 997 1005 10.1002/ca.23249 30168609 17. Maresky HS Oikonomou A Ali I Ditkofsky N Pakkal M Ballyk B Virtual reality and cardiac anatomy: exploring immersive three-dimensional cardiac imaging, a pilot study in undergraduate medical anatomy education Clin Anat 2019 32 238 243 10.1002/ca.23292 30295333 18. Mogali SR Vallabhajosyula R Ng CH Lim D Ang ET Abrahams P Scan and learn: quick response code enabled museum for mobile learning of anatomy and pathology Anat Sci Educ 2019 12 664 672 10.1002/ase.1848 30548228 19. Moro C Štromberga Z Raikos A Stirling A The effectiveness of virtual and augmented reality in health sciences and medical anatomy Anat Sci Educ 2017 10 549 559 10.1002/ase.1696 28419750 20. Kurzweil D Marcellas K Henry B Meyer E Evidence-based guidelines for recording slide-based lectures Med Sci Educator 2020 30 1611 1616 10.1007/s40670-020-01032-w 21. Bolatov AK, Gabbasova AM, Baikanova RK, Igenbayeva BB, Pavalkis D. Online or blended learning: the COVID-19 pandemic and first-year medical students’ academic motivation. Medical Science Educator. Springer US; 2022;32:221–8. 22. Virumbrales M, Elorduy M, Graell M, Mezquita P, Brotons P, Balaguer A. COVID-19: making the best out of a forced transition to online medical teaching. A mixed methods study. MedSciEduc. Springer US; 2022. 23. Iwanaga J, Loukas M, Dumont AS, Tubbs RS. A review of anatomy education during and after the COVID-19 pandemic: revisiting traditional and modern methods to achieve future innovation. Clin Anat. 2021;34:108–14. 24. Pather N Blyth P Chapman JA Dayal MR Flack NAMS Fogg QA Forced disruption of anatomy education in Australia and New Zealand: an acute response to the COVID-19 pandemic Anat Sci Educ 2020 13 284 300 10.1002/ase.1968 32306555 25. Fatani TH Student satisfaction with videoconferencing teaching quality during the COVID-19 pandemic BMC Med Educ 2020 20 1 8 10.1186/s12909-020-02310-2 26. Evans DJR Bay BH Wilson TD Smith CF Lachman N Pawlina W Going virtual to support anatomy education: a STOPGAP in the midst of the COVID-19 pandemic Anat Sci Educ 2020 13 279 283 10.1002/ase.1963 32277598 27. Longhurst GJ Stone DM Dulohery K Scully D Campbell T Smith CF Strength, Weakness, Opportunity, Threat (SWOT) Analysis of the adaptations to anatomical education in the United Kingdom and Republic of Ireland in response to the COVID-19 pandemic Anat Sci Educ 2020 13 301 311 10.1002/ase.1967 32306550 28. Deci EL Ryan RM Vallerand RJ Pelletier LG Motivation and education: the self-determination perspective Educ Psychol 1991 26 325 346 10.1080/00461520.1991.9653137 29. Feri R Soemantri D Jusuf A The relationship between autonomous motivation and autonomy support in medical students’ academic achievement Int J Med Educ 2016 7 417 423 10.5116/ijme.5843.1097 28035054 30. Franchi T The impact of the COVID-19 pandemic on current anatomy education and future careers: a student’s perspective Anat Sci Educ 2020 13 312 315 10.1002/ase.1966 32301588 31. Srinivasan DK Medical students’ perceptions and an anatomy teacher’s personal experience using an e-learning platform for tutorials during the COVID-19 crisis Anat Sci Educ 2020 13 318 319 10.1002/ase.1970 32374937 32. Keller JM. Motivational design for learning and performance: the ARCS model approach. 1 st Ed. Motivational design for learning and performance: the ARCS model approach. New York: Springer; 2010. 33. Huang B Hew KF Measuring learners’ motivation level in massive open online courses Int J Inf Educ Technol 2016 6 759 764 34. Cardoso-Júnior A Garcia VCS Coelho DV Said CC Strapasson ACP Resende IS Translation and transcultural adaptation of the Instructional Materials Motivation Survey (IMMS) to Brazilian Portuguese Rev Bras Educ Méd 2020 44 1 10 35. Cardoso-Júnior A. Evaluation study of motivation in relation to human anatomy video lectures: psychometric validation and application of the Brazilian version of the Instructional Materials Motivation Survey (IMMS-BRV) to medical students. Federal University of Minas Gerais (UFMG): Belo Horizonte, Brazil. Doctorate of Philosophy Thesis; 2021. 36. Cardoso-Júnior A Faria RMD Psychometric assessment of the Instructional Materials Motivation Survey (IMMS) instrument in a remote learning environment Rev Bras Educ Méd 2021 45 4 e197 37. Sbayeh A Qaedi Choo MA Quane KA Finucane P McGrath D O’Flynn S Relevance of anatomy to medical education and clinical practice: perspectives of medical students, clinicians, and educators Perspect Med Educ 2016 5 338 346 10.1007/s40037-016-0310-4 27785729 38. Abdel Meguid EM Khalil MK Measuring medical students’ motivation to learning anatomy by cadaveric dissection Anat Sci Educ 2017 10 363 371 10.1002/ase.1669 27925681 39. Diaz CM, Linden K, Solyali V. Novel and innovative approaches to teaching human anatomy classes in an online environment during a pandemic. Medical Science Educator [Internet]. Springer US; 2021;31:1703–13. 10.1007/s40670-021-01363-2. 40. Singh A Min AKK Digital lectures for learning gross anatomy: a study of their efficacy Korean J Med Educ 2017 29 27 32 10.3946/kjme.2017.50 28264551 41. Barry DS Marzouk F Chulak-Oglu K Bennett D Tierney P O’Keeffe GW Anatomy education for the YouTube generation Anat Sci Educ 2016 9 90 96 10.1002/ase.1550 26061143 42. Keller JM. Motivational design for learning and performance: the ARCS model approach. Motivational design for learning and performance: the ARCS model approach. Springer US; 2010. 43. Choe RC Scuric Z Eshkol E Cruser S Arndt A Cox R Student satisfaction and learning outcomes in asynchronous online lecture videos CBE Life Sci Educ 2019 18 1 14 10.1187/cbe.18-08-0171 44. Meece JL Glienke BB Burg S Gender and motivation J Sch Psychol 2006 44 351 373 10.1016/j.jsp.2006.04.004 45. Bećirović S The relationship between gender, motivation and achievement in learning English as a foreign language Eur J Contemp Educ 2017 6 210 220 46. Hu Y. Motivation, usability and their interrelationships in a self-paced online learning environment. Virginia State University: Blacksburg, VA. Doctorate of Philosophy Dissertation; 2008. 47. Ozan O Ozarslan Y Video lecture watching behaviors of learners in online courses Educ Media Int 2016 53 27 41 10.1080/09523987.2016.1189255 48. Cuschieri S Calleja AJ Spotlight on the shift to remote anatomical teaching during COVID-19 pandemic: perspectives and experiences from the University of Malta Anat Sci Educ 2020 13 671 679 10.1002/ase.2020 32956579 49. Abdulghani HM Sattar K Ahmad T Akram A Association of COVID-19 pandemic with undergraduate medical students’ perceived stress and coping Psychol Res Behav Manag 2020 13 871 881 10.2147/PRBM.S276938 33154682 50. Singal A, Bansal A, Chaudhary P, Singh H, Patra A. Anatomy education of medical and dental students during COVID-19 pandemic: a reality check. Surgical and Radiologic Anatomy. Springer Paris; 2020. 51. Trelease RB From chalkboard, slides, and paper to e-learning: how computing technologies have transformed anatomical sciences education Anat Sci Educ 2016 9 583 602 10.1002/ase.1620 27163170 52. Kurniawan R Zainuddin Z Ishak T The role of pre-class asynchronous online video lectures in flipped-class instruction: identifying students perceived need satisfaction J Pedagogical Sociol Psychol 2020 1 1 11 53. Fleagle TR Borcherding NC Harris J Hoffmann DS Biology C Roy J HHS Public Access 2019 11 385 396 54. Mishall PL, Meguid EMA, Khalil MK, Lee LMJ. Transition to effective online anatomical sciences teaching and assessments in the pandemic era of COVID-19 should be evidence-based. Medical Science Educator. Springer US; 2022;32:247–54. 55. Zalts R Green N Tackett S Lubin R The association between medical students’ motivation with learning environment, perceived academic rank, and burnout Int J Med Educ 2021 12 25 30 10.5116/ijme.5ff9.bf5c 33513127

0

PMC9753862

NO-CC CODE

2022-12-16 23:26:19

no

Med Sci Educ. 2022 Dec 15;:1-10

utf-8

Med Sci Educ

10.1007/s40670-022-01714-7

oa_other

==== Front HEC Forum HEC Forum Hec Forum 0956-2737 1572-8498 Springer Netherlands Dordrecht 36520271 9502 10.1007/s10730-022-09502-x Article Democratizing Conscientious Refusal in Healthcare http://orcid.org/0000-0003-0898-3796 Scott David C. Dscott3@bellarmine.edu grid.252934.b 0000 0004 0429 1132 Bellarmine University, 2001 Newburg Rd, Louisville, KY 40205 USA 15 12 2022 131 18 11 2022 © The Author(s), under exclusive licence to Springer Nature B.V. 2022, Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. Settling the debate over conscientious refusal (CR) in liberal democracies requires us to develop a conception of the healthcare provider’s moral role. Because CR claims and resulting policy changes take place in specific sociopolitical contexts with unique histories and diverse polities, the method we use for deriving the healthcare norms should itself be a democratic, context-dependent inquiry. To this end, I begin by describing some prerequisites—which I call publicity conditions—for any democratic account of healthcare norms that conflict or jibe with CR. Next, drawing on Ronald Dworkin’s jurisprudence and Tom Beauchamp & James Childress’s approach to bioethical reasoning, I briefly introduce one method for generating healthcare norms that is faithful to the publicity conditions and has potential to constructively, and democratically, derive important boundaries for CR. Finally, I argue that many critics of CR fail to similarly ground their accounts of healthcare norms in healthcare professionals’ sociopolitical contexts, often relying instead on their own interpretation of a generally stateable healthcare norm. This leads to their misconstruing both the value judgments on which their own approaches rest and the public, political values that are often invoked in favor of CR. Keywords Bioethics Political philosophy Conscientious refusal Conscientious objection Medical ethics Philosophy of Law ==== Body pmcAs a foreseeable result of the rapidly changing technological possibilities, social conditions, and legal landscapes surrounding healthcare systems, conscientious objection or conscientious refusal (“CR” hereafter) has generated considerable interest over the last two decades. CR occurs when a healthcare provider refuses, for moral or religious reasons, to provide a healthcare service that is (a) within the scope of his or her competence, (b) legally permitted, (c) desired by the patient, and (d) accepted as permissible by a substantial portion of the relevant medical community. Physician-assisted dying (PAD), for instance, has been legalized in some U.S. jurisdictions and recognized as a constitutional right in Canada. And more recently, the United States Supreme Court’s decision in Dobbs v. Jackson Women’s Health Organization, 597 U.S. (2022) leaves American citizens in a state of uncertainty about the legality (and availability) of abortion-related services throughout the nation. For each of these services, physicians disagree over whether taking part in them is obligatory, permissible, or immoral, given their various understandings of their professional, moral roles. Some providers understand the preservation of life to be their non-abrogable duty and decline to participate in PAD or abortion, while others take the intentional alleviation of suffering or the safeguarding of bodily autonomy to be central (or, at the very least, not antithetical) to their role. To provide a more prospective example, gene editing through technologies like CRISPR/Cas9 blurs the line between treatment and enhancement, inviting debates about whether the latter is an appropriate (or obligatory) medical practice. The question that CR presents in each instance is what we should do about these disagreements: allow providers to decline to participate in some services or else require them, upon a patient’s request, to offer the full array of services that have been legally and professionally recognized. Answering this question, in turn, depends on our own conclusions about the moral aims of medicine and the consequent moral duties of individual providers to the public. Thus, the first task in constructing an argument for or against CR is developing a conception of the healthcare provider’s moral role. In much of the existing literature on CR, an argument proceeds in a way that seems appropriate to this task by: (a) Identifying one or more healthcare norms that are central to the provider’s moral role, and (b) Demonstrating why CR is either prohibited or permissible in light of those norms. Mark Wicclair (2011) has offered the following labels for the three types of answers that have taken shape: (1) Conscience Absolutism (“Ab”), proponents of which hold that providers ought to have virtually unfettered liberty to sincere CR invocations of CR, (2) The Incompatibility Thesis (“IT”), so-called because they find CR always or almost always incompatible with the healthcare provider’s role, and (3) The Compromise Approach, which includes various positions that endorse some restrictions on CR.1 Conscience absolutism does not have many proponents in the philosophical and medical literature, but it has been represented in various legislative and regulatory measures in the United States.2 Its proponents place a high premium on the value of moral and religious liberty among health care professionals and cite the long-existing liberties that professionals have in choosing who to accept as patients and which services to offer. While the precise rationale for IT varies, the following norms are often cited in support of influential IT arguments: duties inherent in medical practice to put the patient’s interests first (i.e., above one’s private moral interests) or foster the patient’s autonomy (by maximizing her healthcare choices),3 fidelity to a public oath to provide the full array of services within one’s competence,4 or the paramount social value of a predictable and stable healthcare system (which is disrupted by widespread CR and the vagaries of individual professionals’ consciences).5 In short, there is some determinate healthcare norm that the professional violates by imposing, through his or her CR, a tyranny of his or her subjective moral beliefs on the public or the individual patient. A shortcoming of both IT and Ab accounts is that they omit or inadequately develop their method for deriving healthcare norms that conflict or jibe with CR. Such a method is necessary because debates and resulting policy changes about CR occur in a particular democratic context with its own history, healthcare system, and diverse polity. Any easily stateable healthcare principle (such as “respecting the patient’s autonomy” or “protecting professional integrity”) is subject to different interpretations and in tension with other widely recognized values in healthcare practice. Moreover, a healthcare principle’s meaning and weight might look different across areas of practice, types of medical interventions, and other circumstances. Thus, an account of the healthcare professional’s role that purports to be democratic—rather than merely an extension of the author’s worldview—must show its work; that is, such an account should show how its conclusions about healthcare obligations are rooted in its democratic milieu and how it resolves tensions between widely recognized values. Since IT and Ab accounts often trivialize the challenge of showing their work in this way, they end up resembling the sort of ahistorical (and undemocratic) approach that they ascribe to their opponents, insofar as they simply impose their conception of “good” healthcare on a morally diverse polity. One significant consequence of this is that these accounts often misunderstand (and misrepresent) the political value of concepts often cited in favor of (or against) CR. My aim in this paper is to cast doubt on the possibility of any democratic account of healthcare norms (i.e., one that “shows it works” in the way I have suggested) yielding a context-independent and generalized answer regarding CR.6 This is, incidentally, a defense of the compromise position, which holds that some allowance of CR is the best means of balancing the competing interests implicated by CR.7 My critique of opposing positions will primarily focus on a family of IT accounts that fail to democratize healthcare norms, but only because this position is much better articulated and represented in the CR literature than Ab is. Nonetheless, the kind of methodological shortcoming this paper aims to correct is neither universal among IT approaches nor unique to IT: until recently, relatively little work on either side of the CR conversation has thoroughly engaged resources from democratic theory and jurisprudence in developing an account of healthcare’s normative structure.8 Thus, in addition to providing an account of what it takes to democratize healthcare norms, and showing why accounts from all three types of answers to the CR question fail to do so, I demonstrate the necessity of such resources for having fruitful conversations about the moral aims of medicine and CR. Section “The Publicity Conditions” offers a set of conditions that any (meta)normative account of CR must satisfy, while  section “Beyond the Publicity Conditions: A Constructive Approach to Deriving Healthcare Norms” will offer one such account, which draws heavily from Ronald Dworkin’s theory of jurisprudence. The institutional mechanism for employing such a framework is beyond the focus of the paper, though I suspect it is consistent with such proposals as the licensing board solution from Lynch (2008) or the Uber Conscientious Objection in Medicine Committee proposed by Ben-Moshe (2021). Section “Why Influential IT Accounts Fail to Democratize Healthcare Norms” will then explain why some influential IT accounts fail to democratize healthcare norms pursuant to the constraints outlined in section“The Publicity Conditions”.9 The Publicity Conditions One point of ostensible, widespread agreement in the CR debate is that healthcare’s normative content should be constituted by and reflective of the democratic system to which it belongs.10 Both sides of the debate, however, begin with different conceptions of what it takes to reflect (or flout) the background democratic culture. IT accounts often assume that reflecting the public will is primarily about maximizing access to legalized medical services (often on the basis of the physician’s fiduciary duty or fidelity to their public oath); the public will is flouted where such options are narrowed by the idiosyncratic beliefs of an essentially oligarchic professional class.11 Some compromise or Ab accounts, by contrast, presume (to differing degrees) that one necessary condition of reflecting the public culture is some degree of equal representation and freedom of conscience among the ranks of the professional class.12 While both sides argue that the other fails to accord sufficient weight to a healthcare value (e.g., patient autonomy or the professional’s freedom of conscience), few accounts thoroughly ground their preferred value’s priority in a healthcare system’s democratic milieu. In other words, despite wide recognition that healthcare values are derived from democratic considerations outside of healthcare, most commentators spend inadequate time engaging one another at this level.13 As such, each commentator presumes their own conception of democratic faithfulness and talks past their interlocuters. Fortunately, there is a wealth of mostly untapped resources from democratic theory and jurisprudence, which recent CR work has begun to engage, that might help remedy this conversational impasse. This section will provide a detailed explanation of what it takes for an account of healthcare norms to reflect the liberal democratic milieu (or fail to do so). In other words, I will provide an account of what it means to democratize healthcare norms relevant to CR. First, I identify relatively uncontroversial moral and epistemic features of liberal democracies. Next, drawing on resources from bioethics and democratic theory, I will offer some minimal requirements—which I will call publicity conditions—for any thoroughly democratic account of healthcare norms and CR; those approaches that cannot satisfy one or more of these conditions ought to be ruled out. Finally, drawing on resources from Ronald Dworkin and Tom Beauchamp and James Childress (“B&C” hereafter), I will sketch one approach that passes muster under the publicity conditions and offers some further boundaries on the use or restriction of CR. An especially important sociological feature of liberal democratic systems for our purposes is what John Rawls called the “fact of reasonable pluralism,” according to which a diverse set of religious, moral, and philosophical doctrines is a “permanent feature of the public culture of democracy” (Rawls, 2005, 36).14 These doctrines often stand at the core of individual citizens’ identities and life plans. The explanation Rawls gave for this persistent doctrinal diversity is the “burdens of judgment,” which are the “many hazards involved in the correct (and conscientious) exercise of our powers of reason and judgment in the ordinary course of political life” (Rawls, 2005, 56). If indeed religious and moral pluralism is an enduring feature of liberal democracies, then we would do well to understand our most important social practices (such as healthcare) in light of this condition. Put otherwise, assuming Rawls is correct about reasonable pluralism—and I am unaware of any CR accounts that argue to the contrary—part of our task is to resolve the CR debate in a way that is cognizant of it. Precisely how we should account for our compatriots’ plural worldviews is answered, in part, by a central moral feature of liberal democracies: an understanding of all citizens as free and equal moral agents or, at the very least, a willingness to treat them that way for the purposes of law and public policy. If our compatriot is a moral equal, and her life plan features identity-conferring moral beliefs and projects, then these facts about her should affect how we pass laws and enact public policies regarding our social institutions (or at least those institutions that affect her interests). This does not mean, of course, that our compatriot is always owed the public policy outcome she wants; this would often result in our privileging her priorities over those of other citizens. What it must mean, however, is that public policy outcomes should generally aim to do one or more of the following: (a) Be supportable by reasons that are intelligible to her, (b) Make reasonable accommodations for her view where doing so is feasible, or (c) Include a process that is otherwise fair and permits her voice to be represented. If a public policy process is devoid of such features, then the outcomes it produces would seem to simply reflect our own worldviews entirely at her expense. It would therefore seem to “browbeat” (Gaus, 1996,  123) her or treat her as a mere subject of legislation (see Talisse, 2009, 64) rather than an equal moral agent with a life plan of her own. Insofar as this last claim invites any controversy, it is worth briefly reciting three reasons for accepting it. First, the assumption of equality is internal to the idea of a liberal democracy; to borrow another Rawlsian term, it is a “considered conviction” to which any tenable public policy proposal in a liberal democracy must conform.15 Second, a sincere commitment to democracy includes a desire to keep dialogues of public importance open which, as I suggest later, makes reasonable moral and religious pluralism within our social institutions necessary. If we are committed to truth-seeking in public life, and have sufficient epistemic humility to recognize that we are not the last word on controversial public debates, then certain of our most important practices should leave open the possibility that the moral majority is wrong after all. Indeed, in the tradition of J.S. Mill, much recent work in “new diversity theory” demonstrates the myriad ways in which all of society benefits from such diversity in our social institutions.16 As Roger Trigg (2017) has similarly argued in the CR context, our democratic commitments recommend the nourishment, not the permanent expulsion, of the moral minority on certain controversial medical practices. Third, reasonable citizens are interested in keeping beneficial social practices, like healthcare, stable and thriving. If we again assume that the fact of moral and religious pluralism is a permanent feature of liberal democracies, then our social institutions should be structured (and justifiable) such that they enjoy the support and reflection of a wide swath of the reasonable citizenry. Otherwise, we risk fomenting divisions that can destabilize these mutually and enormously beneficial social practices. In sum, we have moral, epistemic, and prudential (i.e., self-interested) reasons for public policy determinations that accord some weight to our compatriots’ interests, even where we find them wrong or misguided. These general reflections on liberal democracies are instructive in fleshing out the democratic spirit in which we now undertake the task of deriving healthcare norms that are relevant to CR. Our aim is to construe “good” healthcare in a way that does not simply presume our own ideological orientation, but attempts to both argue from shared values and grant a hearing to others’ perspectives. At the same time, our manner of deriving healthcare values must be sufficiently rich that it is useful to the CR conversation. In other words, it must be able to specify what is obligatory, laudable, permissible, ignoble, or prohibited in professional conduct, especially insofar as these judgments reveal what we should do about various instances of CR. The reason we first needed to characterize the “spirit” in which we should construe healthcare norms is that such norms within liberal democracies are not just derived from disparate sources, but are also often vague, incomplete, and in tension with one another. While professional codes of ethics and statutes governing healthcare are perhaps the most easily identifiable sources of professional obligations within a given context, such obligations are also derived from, to different degrees, regulations and other administrative acts, judicial precedent, institutional policies and procedures, hospital ethics committees, democratic principles, and even public opinion. Thus, with multiple sources and interpretations of healthcare norms available to us in many cases, we need a heuristic that ensures we are not just selecting the interpretation among these that aligns with our own worldview. Another way of framing this is that, where a healthcare norm is indeterminate in its meaning or its importance, we need a way of both specifying and balancing the varied and pertinent norms. For example, a norm that is undeniably central in all democratic healthcare systems, like “do no harm,” is often ambiguous in meaning and scope. Thus, when applying it to a particular case, we must first engage in specification, which on B&C’s account means “reducing the indeterminacy of abstract norms and generating rules with action-guiding content” (Beauchamp and Childress 2013, p. 20). As B&C suggest, the generally stateable concept of “doing no harm” is, by itself, useless in determining the permissibility of physician-assisted dying. Even where a norm is sufficiently determinate, there often remains a tension between it and other applicable norms, as relatively few healthcare norms are absolute. This second difficulty requires balancing, pursuant to which we consider the strengths and weaknesses of each pertinent norm in the case before us, resulting in a judgment about which norm should prevail, and to what extent. Cases before hospital ethics committees often weigh the relative importance of respecting patient (or parental) autonomy versus protecting the patient from harm. In order to ensure that balancing involves an unbiased and rigorous reasoning process, as I have suggested it must in liberal democracies, B&C propose a variety of constraints on how balancing can take place in a fair manner, some of which inform the publicity conditions I have set out below. To recap our task as we have summarized it so far: Given the fact of reasonable pluralism among free and equal citizens—together with the vague, disparately sourced, and often conflicting healthcare norms—we need a fair decision procedure for identifying relevant healthcare norms from our milieu and interpreting them in a way that tells us what to do in specific cases of CR. While this process will draw from principles within healthcare, these principles are themselves determined by public priorities outside of healthcare. Put otherwise, our construction of the “good” of healthcare requires the consideration of sources and principles that are facially non-medical. Pursuant to these observations, I offer four, minimal conditions that any account of healthcare norms must meet in liberal democracies. These conditions follow from features of liberal democracies I have identified so far and from principles of good argument and epistemic humility. In a Rawlsian vein, I will call these publicity conditions, though I intend them to be consistent with more models of justification than the Rawlsian idea of public reason. These conditions are not sufficient to construct a full account of healthcare obligations by themselves, but they are fruitful in identifying, and ruling out, undemocratic accounts. First, we cannot build an assessment of religious claims or of conscientious objectors into the test we use to evaluate them. This might come about if one were to suggest that “a willingness to put one’s conscience aside” is the non-negotiable healthcare virtue that CR violates. Since the whole point of the CR debate is to demonstrate the propriety or impropriety of conscience-based exemptions, and whether one should put one’s conscience aside, we cannot simply exclude it by definitional fiat. In other words, the claim that “putting one’s conscience aside” is a virtue must be justified by reference to other, shared values (i.e., shared with those who doubt whether it is a virtue). Nor will it do to define medicine’s normative structure in a way that depends on a specific attitude toward religious professionals or religion itself, for that would also be to similarly beg the question about which attitudes belong in healthcare professions. After all, the consequence of debating CR in a liberal democracy—given the fact of reasonable pluralism—is that we are tasked with making our case intelligible to those with radically different religious attitudes than us.17 What these observations suggest, to put the first publicity condition succinctly, is that we must offer independent reasons for CR’s propriety or impropriety. This first condition is like the idea of public reason that appears throughout the Rawlsian tradition, pursuant to which coercive state action should generally be supportable by reasons that all reasonable citizens could be expected to accept.18 Benjamin Zolf (2019) has argued that a policy allowing CR cannot be justified by public reason because CR claims are generally motivated by religious doctrines that make controversial claims. But this rests on a misunderstanding of how Rawlsian public justification works: the appropriate standard is not whether the religious professional’s worldview is accessible to the public, but whether there are accessible reasons supporting a policy that allows CR. Zolf also claims that the coercive state action that triggers public reason concerns is the licensure of individual professionals who have a particular belief on a controversial matter in healthcare. But if there are non-objecting professionals who that patient might seek out instead, it is not clear how they are coerced. Moreover, on this reasoning, would not the alternative—foreclosing the licensure of those with differing beliefs on a controversial matter in healthcare—be significantly more restrictive? While this is a somewhat tangential point for our discussion, I mention it because it is worth emphasizing that the idea of public reason, insofar as it applies to CR, works in both directions: a policy must be justifiable to all parties whose interests it might affect: patients, prospective and current healthcare workers, and the diverse interests of the citizenry. Second, similar cases (i.e., those involving similar parties or competing interests) should be decided similarly. Many of the arguments against CR depend on the notion that there are potential harms to patients, the public, or the profession that are sufficiently contrary to good medicine that health professionals should avoid conduct that even risks bringing those harms about. Assuming there is such a duty (i.e., to avoid the risk and not just the occurrence of harm), then presumably any act or omission that creates risk, whether or not that act or omission is religiously or morally motivated, should be prohibited or discouraged. Many of the arguments in favor of CR depend on the notion that more religion in the public square is an unequivocally positive development. To borrow some pertinent terms from First Amendment jurisprudence in the United States, democratic norms about the refusal of medical interventions must be viewpoint-neutral and generally applicable, rather than targeting specific worldviews or motivations for refusal. In short, our account must be marked by equal treatment of various citizens’ worldviews. Third, a normative foundation for healthcare, like any social practice, must be traceable to considered convictions that form part of our shared history. The four principles B&C identify—beneficence, non-maleficence, autonomy, and justice—are central to Western medicine, even if (a) we might sometimes disagree about the application and weight of these principles and (b) their relative ordering in our bodies of positive law change over time. Experiments at Auschwitz and Tuskegee provide paradigmatic examples of what is to be avoided when experimenting on human subjects, leading to the development of contemporary research ethics in opposition to those atrocities. Through EMTALA, the United States, like many other countries, guarantees universal access to emergency care regardless of ability to pay, a judgment that I take Paul Menzel to be quite right in deeming a fixed part of our moral fabric. Each profession has long-recognized specific, patient-facing duties—such as maintaining clinical competence, keeping patient confidence, or explaining procedures in an honest yet sympathetic register—about which it is difficult to imagine disagreement. Each of these examples is easily locatable within professional codes of ethics, enshrined in law, or the product of universal consensus among competent practitioners and public stakeholders. An account of healthcare’s normative structure ought to be supportable by principles that are universally shared (or very nearly so) and consistent with the evolutionary path our healthcare system has followed. In the following section, I will provide a framework that further illustrates how this condition might be satisfied. Fourth, a democratically situated healthcare system, though it must have substantive moral priorities and aims, should not be blind to reasonable parties’ competing claims and interests. Recall that we are generally loath to coerce our compatriots for reasons that are inaccessible to them, as doing so treats them as mere subjects of legislation rather than autonomous (and equal) moral agents. In specific circumstances where multiple parties’ putative rights are at stake, we ought to avoid all-or-nothing policies where feasible, i.e., those in which one party is an outright or permanent loser. While political or moral principles often supply reasons to accord more weight to one kind of party (e.g., a patient or pro se litigant) or one kind of claim (e.g., violation of a fundamental right), this does not imply that the less-prioritized party is accorded no weight at all. Even where taking up a profession is a voluntary act, such that placing conditions on entering that profession is less problematic than it is for universally obligatory acts (like military conscription), the former can still be quite problematic. The “reasonableness” qualification in this fourth condition recognizes that we cannot (and should not) include all conceivable perspectives in balancing affected parties’ interests. Our construction of healthcare’s normative structure need not be addressable to unreasonable persons. The unreasonable person, in Rawlsian parlance, is marked by a zeal to embody the whole truth (their version of it) in politics at the expense of others. To borrow an example from Jonathan Quong, we need not make our laws concerning theft intelligible to someone who claims a religious right to steal my computer, as the assertion of such a right contradicts the environment in which equal rights are possible (Quong, 2004, 333). The reasonable person, by contrast, is one who exhibits a willingness to (a) abide by fair terms of cooperation provided that other citizens do so, (b) recognize that citizens of good will might disagree on matters attributable to their worldviews, and (c) make their reasons for their political actions intelligible to others and be willing to listen to others’ reasons and accommodate them where feasible. Not every instance of living in accordance with one’s worldview, however, constitutes this kind of indifference towards others’ moral lives.19 Speaking to the matter at hand, there is a significant difference between the physician who declines to offer a specific service, yet refers potential patients to a physician who does offer it, and one who actively aims to prevent patients from obtaining that service elsewhere. Moreover, the objecting professional, or the Ab proponent who supports her, is not the only one who risks behaving unreasonably, an assumption which is implicit in many IT accounts. The policymaker who wishes to impose an excessive degree of moral uniformity on the healthcare profession, by foreclosing the possibility of dissent or pluralism, is also an unreasonable actor with an anti-democratic motive. Beyond the Publicity Conditions: A Constructive Approach to Deriving Healthcare Norms The publicity conditions are a preliminary test in determining precisely which of our judgments about good medicine could conceivably translate to universal obligations for professionals of a given type. Thus, I have offered them as an initial check on any method of deriving healthcare norms, even if they are too thin to supply such a body of norms by themselves. For a more substantive evaluation of whether various cases of CR fit the professional’s role, we require additional conceptual equipment, some of which I briefly sketch here and fully develop elsewhere. The Dworkinian, two-stage model I propose aims for an account of medical ethics that is internally coherent, fair, and historically grounded. It also does not purport to be the only appropriate model for deriving professional obligations. Thus, while the publicity conditions are offered as prerequisites to any construction of healthcare obligations in liberal democracies, this two-stage model is perhaps one of several conceptions that might satisfy them.20 One of Dworkin’s most influential metaphors—that of the chain novel—might be helpful in providing an overall characterization of the model this section proposes. Dworkin asks us to imagine that Charles Dickens never wrote A Christmas Carol and that we have been furnished with the first part of the novel, with several authors having written these earlier chapters. It is then our task, as author of one or more later chapters, to make it the best novel that it can be, giving it a structure and purpose that it would have if written by a single author. In doing so, we are somewhat constrained by the data we already have before us—the chapters already written—and are cognizant of the guidance we are providing to authors of the chapters after ours. While it is likely the case that there are several ways of interpreting the themes, purposes, or currents of character development taking place in these earlier chapters, our choices are far from limitless if we want to give the novel a high degree of integrity. In general, if we are being faithful authors and literary critics, we will not leave unexplained some “dominant and repeated metaphor” or “a subplot treated as having great dramatic importance” from the preceding chapters (Dworkin, 1986, 230). At the same time, our chapter(s) need not give full effect to every minor element of the preceding chapters: certain elements can be identified as accidental or as mistakes in light of the novel’s overall purpose. Thus, if we were tasked with picking up the novel after Scrooge has already been visited by the three spirits and repented for his past callousness and cruelty, it would make little sense to begin the next chapter with the revelation that Scrooge is irredeemably wicked after all; the only interpretations available to us would seem to be ones that set Scrooge on a path to redemption of some sort. If “our” authored chapter took place earlier in the novel, however, then this interpretation of Scrooge’s character might still be available to us. In short, our job is neither a wholly mechanical one nor is it one in which we are to begin a new novel of our own: we are authoring a chapter that is the result of our faithful interpretation of the novel as a whole. Insofar as two interpretations are available to us, each of which render the text coherent and purposeful, we might then engage our own, substantive view about what makes for a good novel in choosing between those two. Just as Dworkin suggests that the chain novel provides a fitting metaphor for the tasks of judges interpreting the law, so too does this section suggest it provides a fruitful way of constructing the normative structure of healthcare professions. When we ask whether a particular medical intervention is obligatory, or when we ask what the obligations of healthcare professionals are (perhaps in order to answer the first question), we are inevitably engaging in an interpretive process. Some rather particular obligations are set in stone or locatable in a singular source (e.g., a federal law), while others are more diffuse or amorphous, as recurring “themes” in our story. Where two different conclusions about a healthcare practice both explain the elements and direction of a healthcare system, we might then consult our own moral and political ideals in choosing between these two conclusions. In short, as this section will explain in greater detail, we want to tell a story about healthcare norms that is both faithful to the direction of the novel as told by its authors so far (healthcare professionals, political actors, citizens, patients, etc.) and that aims to make the healthcare profession(s) the “best” kinds of practices they can be. Following Dworkin, I will call the first of these stages “fitness” and the second “justification” (which Dworkin also refers to as “substance”). Stage 1: Fitness The first step in this method is to determine, in a way that builds on the work of the third publicity condition (a connection to our “shared history”), whether a proposed rule governing an instance of CR coheres or fits with the physician’s (or other professional’s) role. This determination, as Dworkin said of adjudication, is an interpretive act that aims to resolve a theoretical disagreement about medical practice. We ostensibly agree about the ultimate referent of the CR conversation—medical practice within liberal democracies—but disagree about what that practice requires. As I suggested earlier, this disagreement is not only a consequence of our conflicting worldviews, but the disparate and diverse sources from which we derive professional obligations. A proposed rule demonstrates fitness where it generally reflects, rather than conflicts with, the pertinent parts of these various sources. The determination of fitness necessarily takes place at multiple levels of specificity, as each jurisdiction includes norms internal to democratic citizenship, healthcare generally, specific healthcare professions, and certain specialties or sub-specialties within those professions. Fitness does not require a proposed rule to explain every one of these sources, some of which might be anachronistic, more aberrant,21 less binding, or more controversial than others. But we ought to look with suspicion upon a putative norm that conflicts with or fails to explain several major sources of obligations in a given sociopolitical context, or that makes little attempt to reconcile existing sources of norms that are in a small degree of resolvable tension with one another. A rule that departs too much or too often from these sources is not an attempt to describe medical obligations as they are currently understood, but an attempt to describe an essentially new practice or impose one’s own conception of the good.22 Fitness is not only desirable for obvious democratic reasons, but also because we wish for a principled, purposive basis for our most important social practices, as opposed to what Dworkin refers to as arbitrary “checkerboard” solutions.23 Suppose we are determining what an OBGYN’s duty is when it comes to either a long-legalized practice, like abortion, or a prospective one like genetic enhancement (in a world where services related to CRISPR are the province of OBGYNs). We might imagine a list of possible norms that describe the OBGYN’s obligation, including norms that take abortion or genetic enhancement to be: exceptionless obligations, generally applicable obligations with limited exceptions, permissible and non-obligatory services, or generally prohibited. When evaluating these candidate norms for fitness, we might begin with and give preference to localized considerations, as these bear directly on the question of what might be expected of individual physicians, as opposed to generalizing about all practitioners in all contexts. If the OBGYN specializes in maternal-fetal medicine (MFM) in the United States, we might look to the position statements and ethical guidance from the Society for Maternal-Fetal Medicine and any licensing or regulatory bodies that oversee MFM. Then, to a progressively lesser (but still significant) extent, we might turn to professional organizations overseeing all OBGYNs, all physicians, and perhaps even other, similar healthcare providers. Parallel guidance from other healthcare systems, especially liberal democratic ones, also deserves consideration as persuasive authority. This sort of excavation will inevitably produce ambiguities, including the relative weight of certain sources of healthcare norms, that would need to be resolved through whatever administrative mechanism or quasi-judicial setting this procedure runs. For instance, how might we determine the content of certain core duties that apply to all American physicians?24 Counting an American Medical Association’s position statement on an issue as being dispositive would exclude the voices of the overwhelming majority of American physicians, who are non-members and not otherwise bound or represented by the AMA’s views. On the other hand, turning to a simple majority of individual physicians’ practices would be inadequate for several reasons. First, it’s not clear that we can glean providers’ positions on certain types of interventions based on their provision or non-provision of that service: they might not offer a certain intervention out of a lack of interest or they might offer a service without deeming their differently-minded colleagues negligent for not offering it. Second, treating the practices or opinions of a simple majority of practitioners as dispositive would somewhat vitiate one of the oft-cited reasons for CR: ensuring that minority voices are not automatically barred from certain professional opportunities. Moreover, even where there is an obvious consonance between a professional organization’s position statements and a supermajority of practitioners, scenarios are foreseeable in which both sources differ dramatically from the expectations of a supermajority of the citizenry. While it is outside the scope of this essay to address the various iterations of such scenarios, their inevitably comes with several implications for actually applying this methodology. A provider using this methodology as a guide for their own conduct might simply (albeit in good faith) consult their sense of what their profession’s consensus (or lack thereof) is. Employing this methodology to determine public policy, however, will require an adjudicative or quasi-adjudicative body that is actually in a position to examine the evidence from various sources and assign these sources weight in a non-biased manner. Fortunately, this task is far from foreign to policymaking in healthcare: much of medical malpractice law relies on the testimony of experts about the standard of care within various disciplines. Similarly, it might be that a committee—like Ben-Moshe’s UBCOM Committee or Lynch’s expanded role for licensing boards—setting CR policy ought to be comprised of representatives from the pertinent professional association, non-member professionals, patient advocates, and other stakeholders. In cases of consonance between a professional association (like the AMA) and the bulk of practicing physicians (as determined by an expert witness or representative) about the obligatory status of a particular medical intervention, then an especially strong showing from outside sources (e.g., patient or citizen advocates) would be required to dislodge the presumption in favor of its obligatory nature. Cases of dissonance might require the adjudicative body to look elsewhere (i.e., to more general principles and practices) for guidance in the fitness analysis. In any event, the aim of “preferring local” sources is to construct an account of what is expected of this sort of practitioner in this sort of context. Only a particularized account like this, as mentioned earlier, can indicate which practices are core, penumbral, central, peripheral, permissible, ignoble, prohibited, or laudable within that area of practice. These vital distinctions are typically absent from most IT or Ab accounts and, insofar as we want to insulate our accounts against arbitrariness and bias, we want to ground them in socially locatable sources.25 Moreover, we generally have good reason to suppose that those on the inside of a profession have knowledge and experience that warrants some degree of deference regarding services that are beyond public understanding or regular public concern.26 As such, insiders’ voices are weighty in the fitness analysis without being dispositive or insurmountable. Since the presumption for “local” considerations has a specific, democratic purpose, it is conditional and has limits. First, the “local” rule or convention might be so aberrant against the backdrop of less localized sources that the presumption in favor of the local is rebuttable. There may be special interests that afflict a professional organization or government entity, archaic rules that no one has bothered to revisit or correct as public morality has otherwise shifted, or other reasons why less localized sources offer better evidence of how society views that profession’s role. Second, a source can enjoy the same weight as a transparently local consideration where it is clearly intended to apply to all physicians, professionals, or citizens. Third, notwithstanding my use of codes of ethics or regulations as examples, these need not always enjoy preferred status over less explicitly codified norms from the political or medical milieu, especially where the latter are firmly established. The valorization of good bedside manner and admonition against patient abandonment, for instance, would seem to be on equal footing with various, more facially local sources. The specification of patient abandonment (i.e., which actions constitute abandonment), however, might look different in various specialties. Fourth, less localized considerations might be used where the more localized source is silent on the issue at hand, or where a particular understanding of the localized consideration puts it into a high degree of tension with other normative sources. Overall, the idea of local preference should reflect both the morally diverse public’s considered convictions, on the one hand, and the informed application of those convictions by those internal to each specialty, on the other. The “preferring local” qualification, insofar as it aims to reflect the democratic culture in which medical practice takes place, necessarily includes a temporal component: what I will call the “preferring recent” qualification. Recent normative shifts—or rather, sources of normative obligation that are more recent in time—are generally weightier in the fitness analysis than contrary and preexisting normative priorities, common medical practices, professional rules, etc. As an obvious application of this “preferring recent” qualification, pre-1970s attitudes towards patient autonomy are clearly of much lesser weight, and perhaps no weight at all, given the myriad legislative, judicial, and professional changes of landscape that make patient autonomy central to contemporary healthcare. Sources of this shift in healthcare include not just well-publicized documents or landmark legal developments (like the Belmont Report of 1978, the Patient Self-Determination Act of 1990, HIPAA, Roe v. Wade, etc.), but numerous developments in informed consent law, hospital policies and procedures, medical education, and professional codes of ethics. As public policy and professional ethics became more protective of patient autonomy, this entailed a larger set of enforceable obligations on the part of providers. At the same time, technological developments expanded the possible scope of medical interventions beyond traditional and unanimously endorsed conceptions of healing. Consequently, these were the first decades in which a tension between patient and physician autonomy could have arisen, which led to some of the earliest protections for physician conscience. In short, the circumstances of the last several decades have seen provider integrity and patient autonomy both become entrenched in various legal and institutional sources of western healthcare systems. The same is true regarding an example mentioned in section “The Publicity Conditions”: EMTALA’s passage, the various practices that arose around its implementation, and its bipartisan popularity suggest it is much more an ineradicable part of our moral fabric than the preceding centuries in which healthcare systems included no such guarantee. It would be absurd to view these later measures as aberrations or departures from the healthcare story rather than new chapters that were now the main plot. While a thorough fitness inquiry, for either abortion or CRISPR, is beyond the purpose of this paper, I will briefly comment on some likely starting points given the guidelines I have just laid out. A proposed rule for MFM specialists in the United States that would always obligate them to provide either service finds little (if any) any support within the norms and conventions of MFM, obstetrics, medicine, or other, related healthcare professions. Professional organizations and regulatory bodies have either explicitly supported CR (with restrictions) or, in organizations that have not, included general language defending the professional’s moral integrity. In terms of professional practice, an overwhelming majority of OBGYNs do not provide abortions and residency programs commonly accommodate residents’ requests to assist in no or only some abortions. Yet reproductive freedom is also deeply entrenched in the moral and legal framework of Western democracies. Abortion was recognized as a constitutional right in the United States for nearly a half-century. Even though the U.S. Supreme Court’s recent decision in Dobbs v. Jackson seems to have altered the legal landscape, it remains the case that: (1) The long-term legacy of Dobbs is unclear, (2) The Dobbs decision itself does not affect the right of various states to continue providing abortion-related services, and (3) There is opposition to the contrasting frameworks of Dobbs and Roe v. Wade among both the citizenry and medical professionals. This tension seems to come with two implications for a fitness inquiry in a jurisdiction that is relevantly like the United States. First, a general right to an abortion need not imply an absolute obligation for an individual physician; governmental entities or licensing bodies might be the most appropriate bearers of this obligation. Second, this suggests that cases like these simply do not lend themselves to generalized answers like those that IT or Ab proponents offer. In other words, it might be wrongheaded for either side of this debate, at least where a tension between two ostensibly entrenched principles exists, to ask whether there is a right to CR in general or not. A nuanced fitness analysis can also reveal the folly of arguments that make CR an all-or-nothing affair, i.e., those in which either any imaginable conscience claims are admissible or none are. IT proponents often argue that opening the door to any CR claims at all results in a deluge of CR claims that are manifestly contrary to any tenable conception of medical practice. Alberto Giubilini (2017) has argued, for instance, that there is no principled way of distinguishing between a physician who regards the fetus as a person with inviolable rights from one who thinks the same of bacteria. Thus, if we were to recognize the former physician’s right to not provide abortions, so this argument goes, we could only exclude the latter physician’s CR claims on parochial or arbitrary grounds. But Giublini mistakenly assumes that the reason for accommodating certain worldviews has a principally metaphysical (and prejudicial) basis. On this understanding, allowing CR on grounds of fetal personhood, but not on bacterial personhood, would be either arbitrary or else contingent on the greater social capital of pro-fetus advocates. If norms about professional obligation are derived from public moral culture, as I have suggested they are, then Giubilini’s claim falls apart. The use of antibiotics, elimination of illness-causing microbes, and curing of human bodies are all firmly entrenched components of medical practice across specialties and jurisdictions. While the concept of human beneficence evolved in twentieth and twenty-first century medical practice—to prioritize, for instance, mental health and palliative considerations—its pursuit has remained the core of all medical specialties. This is not a matter of mere majority belief, but both a quantitative super-super-majority and qualitative judgment about the sine qua non of healing. Whatever our ultimate judgments about CR, there is a clear difference in terms of fitness, and not simply substantive metaphysics or metaphysical neutrality, between treating the fetus as a patient (or at least a being of some medical priority), and behaving similarly towards bacteria. Among other socially locatable indicators of this difference, specialties devoted to fetal health and an infant’s (legal and medical) status immediately after birth attest to this. There is nothing, on the other hand, that is socially locatable regarding bacterial rights nor a political understanding of toleration that could warrant exposing patients to untreated bacterial infection. By the same token, and even more transparently, a fitness analysis demonstrates the untenability of Ab in the healthcare systems of western democracies, especially the United States. Consider, for instance, a hematologist or critical care professional who declines to participate in blood transfusions. Whatever the value of freedom of conscience or the authority practitioners otherwise have in determining the scope of their practice, there are several reasons that allowing CR in this context would fit very poorly with (for instance) American healthcare’s normative history. First, blood transfusions are often performed in emergency settings where there is insufficient time to find a replacement for the objecting professional. As noted earlier, measures like EMTALA indicate the paramount importance that American healthcare places on guaranteeing emergency care. Second, and relatedly, American law and professional ethics—both within and outside of healthcare—includes myriad admonishments again patient abandonment and negligent rescue. In other words, there generally no duty to perform a rescue (as a citizen) or take on a client-patient (as a professional), but there is a well-recognized duty to see the job through once one has assumed care for that person. In an emergency care setting, there is an ongoing, immediate assumption of care taking place that sets it apart from other healthcare contexts. Third, the notion of wellness at which blood transfusions aim is an uncontroversial and, yet again, super-super-majority one, so much so that it is difficult to imagine clinical hematology or emergency care without it. The failure to expeditiously perform a blood transfusion (and doing so non-negligently), where medically indicated, is commonly deemed an instance of medical malpractice. There are might well be healthcare contexts and patient circumstances in which abortion, or some other oft-contested service, shares one or more features with cases like this (i.e., those in which an instance of CR recognition has very poor fitness). But there are also cases involving a unique set of circumstances in which a provider’s objection to what are ordinarily “core” or commonplace interventions within their specialty or related specialties does fit. Whatever the outcome of particular cases, this fitness inquiry aims at ensuring that providers and policymakers are writing the next, coherent chapter of our healthcare story, and not embarking on a novel that is wholly their own conception. Stage Two: Justification Whatever direction the fitness analysis takes in a particular instance of CR, let us suppose several candidate norms survive a full fitness inquiry. At this point, we exhaust the descriptive account of medical obligation (i.e., the portion of the account that locates moral expectations of physicians and other healthcare professionals in socio-political sources). Thus, we have run out of clearly neutral ground and have little choice but to engage some of our substantive moral convictions in choosing among the remaining candidate norms. The purpose of this second step is to decide which of the remaining candidate norms do the best job of justifying the practice of medicine and making it a practice worth preserving. Judges, Dworkin observes of this second step, are often deciding between different political values—like justice, liberty, and fairness—in deliberating on the appropriate rule to decide the case, though all judges accord some weight to each of these values. One judge will interpret and balance these ideals differently than the next, and most judges will “think that the balance between the opinions of the community and the demands of abstract justice must be struck differently in different kinds of cases” (Dworkin 1986, 250). For our purposes, in considering an instance of CR, we might take ourselves to be policymakers behaving like judges or healthcare professionals deliberating on their own obligations.27 There are many healthcare desiderata at play in moral conflicts, such as B&C’s four healthcare principles or various principles of political morality (non-coercion, equal opportunity, respecting reasonable pluralism, social stability, etc.). After the fitness inquiry in an instance of CR, remaining reasonable disagreements might include conflicts of interpretation of a healthcare principle (such as non-maleficence), conflicts between a healthcare principle (like autonomy) and a principle of political morality (like preserving moral or religious diversity), and conflicts between principles of political morality (a predictable healthcare system versus one that is responsive to a morally diverse polity). This step leaves us some latitude, or what Daniel Sulmasy calls “discretionary space,” to determine precisely how to interpret and balance them in the wake of our fitness inquiry. I will offer two, brief examples that illustrate some limits and features of our reasoning in this second, justificatory stage, one which bears directly on CR and another that might prove analogically useful. Though this stage is prescriptive, it remains constrained by the results of the first stage (fitness) in two ways: (1) The eventual rule must come from among the norms that survive the fitness inquiry, and (2) The rule should be justifiable in a way that is intelligible to the affected parties (the current patient, future patients or patients elsewhere, healthcare professionals, the public, etc.). It would therefore remain inappropriate (in nearly all imaginable circumstances) to engage one’s convictions in a way that justifies compelling treatment against a medically competent patient’s will, even if my moral judgments about balancing beneficence and patient autonomy warrant divergence from some of my colleagues on other issues. To borrow an example from Wicclair, suppose a family physician or nurse recently converted to Buddhism and believes that “pain is the working out of life’s karma,” (Wicclair, 2011, 93) and decides that interfering with his patient’s physical suffering is in fact a great spiritual harm. These would be inadequate grounds for conscientious objection to pain medication, Wicclair argues (I think correctly), because controlling pain through medication is not simply permissible and accepted, but a core goal of family medicine or nursing, and therefore obligatory.28 There are circumstances involving related, palliative issues—like preserving quality versus quantity of life, the social impact and justice of certain pain management practices (especially those giving rise to opioid addiction), medical futility, etc.—that present tensions and ambiguities among healthcare goals for which moral discretion is appropriate. More generally, there are ongoing controversies, especially in light of growing technological possibilities in medicine, about certain understandings of “harm.” On my account, these circumstances differ from Wicclair’s Buddhist physician not due to moral or metaphysical beliefs I hold, but from the indeterminate results of their fitness (the first stage of analysis) inquiry. Disclosure requirements within clinical trials offer another example in which there are actions that society has emphatically proscribed—such as lying to patients about the purpose of the research29—and others about which there remains disagreement, including details of how to appropriately communicate risks and benefits to clinical trial participants. There remain disputes about the balance between benefits to society (i.e., future patients), on the one hand, and unacceptable risks to current patient-participants (regardless of their consent) or the extent of material disclosures required (to protect patient autonomy), on the other. Wherever this is the case, a fitness inquiry might not always yield a clear answer, such that considerations of political morality, the social purposes of medical research, medicine’s essential virtues, or the value of reasonable pluralism within medicine are appropriate considerations in (a) a physician’s deciding on a rule for her own conduct and (b) an adjudicative body’s deciding between complete uniformity among practitioners versus discretionary choice among clinician-researchers. In sum, there are many cases for which the fitness inquiry determines the obligatory course of conduct for a physician, or at least narrows his or her permissible options. But in those cases where the fitness inquiry is indeterminate, the justification stage leaves us with numerous desiderata to balance in one way or another. What I will suggest in the following section is that some IT adherents take for granted the non-coercive or maximally democratic footing on which their accounts rest, as though they (unlike the CR proponents they criticize) leave their idiosyncratic, personal convictions aside. To the contrary, even assuming IT passes the fitness inquiry in each case of CR, it constitutes only one way of balancing the various healthcare desiderata, and one they offer few public reasons (pursuant to the first publicity condition) to accept over the alternatives. Put otherwise, IT accounts are neither worldview-neutral nor non-coercive to all stakeholders in CR policy. Their assumptions are often that maximal patient access to maximum services and near-total uniformity within the healthcare system outweigh other desiderata. Thus, not only do they leave other desiderata (or competing interpretations of these desiderata) entirely to the side, but they assume that the same balance of them obtains across jurisdictions, specialties, and individual cases. As we will see, neither of those shortcomings bode well for their accounts, insofar as they aim to be historically grounded and democratically responsive. Why Influential IT Accounts Fail to Democratize Healthcare Norms The primary purpose of sections “The Publicity Conditions” and “Beyond the Publicity Conditions: A Constructive Approach to Deriving Healthcare Norms” was to offer some democratic contours on any methodological approach to deriving healthcare norms (section“The Publicity Conditions”) and show how one such approach might yield fruitful answers to the CR debate (section “Beyond the Publicity Conditions: A Constructive Approach to Deriving Healthcare Norms”). A secondary purpose of section “Beyond the Publicity Conditions: A Constructive Approach to Deriving Healthcare Norms” was to cast doubt on the possibility of a democratic approach yielding acontextual or generalized answers to CR. In other words, it is doubtful that such an approach would lead to either the Incompatibility Thesis (IT) or Conscience Absolutism (Ab). This section will more directly demonstrate the ways in which these arguments often run afoul of the publicity conditions outlined in section “The Publicity Conditions”. This section will focus on a certain family of IT accounts not because Ab positions fare any better, but because Ab has virtually no representation in the CR literature. As such, I hope my expeditious dismissal of Ab in section “Beyond the Publicity Conditions: A Constructive Approach to Deriving Healthcare Norms” will suffice so that section “Why Influential IT Accounts Fail to Democratize Healthcare Norms” can better address the more formidable challenge to compromise accounts. Though IT accounts often fail to employ a sufficiently democratic method for deriving healthcare norms relevant to CR, this does not mean that that the concerns which motivate these accounts are undemocratic. To the contrary, the foreseeable consequences of unregulated CR—such as patients being harmed by delayed or unavailable care—should be troubling to any conscientious citizen in a liberal democracy. Indeed, the CR literature (including accounts that are critical of IT) includes many unfortunate instances of such consequences, even in jurisdictions with restrictions on CR. Thus, many IT accounts are successful in both raising significant challenges for any CR-accommodating policy proposal and highlighting restrictions on CR claims that any democracy ought to recognize. Where each of these accounts fall short, however, is in demonstrating that their well-motivated concerns about unrestricted CR require a near-blanket ban on it. There are three broad reasons for this, each of which can be traced to the absence of the sort of inquiry I introduce in sections “The Publicity Conditions”–“Beyond the Publicity Conditions: A Constructive Approach to Deriving Healthcare Norms”. First, IT accounts provide an inadequate appraisal of the value judgments on which their accounts rest, taking their proposals to be value-neutral (as opposed to the parochial reasons that justify CR accommodation). Second, they provide a superficial or inadequate account of the countervailing democratic values that support CR; they often suppose that CR is intelligible only as a self-serving practice. Third, their arguments typically depend on their interpretation of a generally stateable healthcare norm that they fail to locate in the democratic milieu(s) of the pertinent healthcare systems. In this section, I describe three influential arguments against CR before demonstrating how each of them commits one or more of these errors. Fidelity to One’s Public Oath On many IT accounts, CR violates fidelity to the public oath that healthcare professionals take, and on which the public relies, to act as fiduciaries for current and future patients. Upon completing his pre-professional training and entering a healthcare field, the healthcare professional assumes a general “obligation to place the well-being and rights of patients at the center of professional practice” (Stahl and Emanuel, 2017, 1382). Such a commitment includes both duties regarding his interactions with patients (e.g., keeping them informed and engaging them in a compassionate manner) and his clinical behaviors. Regarding the latter, the public expects professionals to maintain a sufficient knowledge base and skillset to discuss and, where they are medically indicated, competently perform the range of interventions that have been legalized and made part of their practice area. In most jurisdictions, there is an oversupply of persons interested in becoming healthcare professionals and willing to assume these duties. Thus, “those who are unwilling to participate in such expected courses of action should not join professions tasked by society with the provision of such services” (Savulescu and Schuklenk, 2017, 165). Those who are aware of such expectations when they join the profession and decline to live up to them when they claim CR, make a “mockery of their graduation promise” (Savulescu and Schuklenk, 2017, 165) and frustrate the public expectations on which their licensure is predicated. While it is difficult to understand healthcare professions in liberal democracies except as a commitment of this general kind, this argument takes the extent or scope of this commitment for granted. IT accounts often assume that an intervention’s being legalized constitutes a moral judgment from the citizenry that its provision is morally important,30 or that its subsequent acceptance by some professionals in a specialty makes it as much a “part” of that domain as the interventions its professionals already offered. Thus, an intervention’s legalization or medical acceptance makes it part of the package of services included in the professional’s “graduation promise.” This assumption is unwarranted for two reasons. First, it is difficult to imagine how a sincere, general promise of this sort is possible in the lived experience of most healthcare workers. Professionals progressively discover what an area of medical practice involves (perhaps including its less savory and surreptitious elements), let alone their appraisal of it, as they advance through different stages of education, training, and professional experience. The physician might only gradually understand the nature of certain interventions or other practices that her colleagues perform, practices which are either very infrequent in her employment context or contrary to the conception of healing patients that her specialty otherwise pursues.31 She might reasonably take her commitment to apply to a breadth of other practices that her colleagues emphasize in their professional development, but not to this one. An IT adherent might reasonably reply that such a practitioner, regardless of her subjective expectations, should have known that she was committing to offering this practice, as she is charged with fulfilling society’s expectations, not her own. But this reply leads to the more fundamental problem with this argument: it often cannot locate the content of this putative promise in the relevant jurisdiction’s milieu. The legalization of a practice is often not a moral approval of that practice, but an expression of society’s political values. Some legalized practices, it is true, also include both (a) the right of any citizen to access that resource or practice and (b) the corresponding obligation of some person or entity to provide it; but neither of these components necessarily follow from the bare fact of legalization. All that legalization of a practice—such as physician-assisted dying or genetic enhancement—clearly represents in a liberal democracy is that the practice is now permissible, and not necessarily obligatory or even encouraged. All that the subsequent performance of a legalized practice among some (or even most) practitioners represents is that this medical intervention does not violate good medicine, but this is distinct from a judgment that this practice’s omission violates good medicine. Indeed, it is commonplace in many jurisdictions to sub-specialize or otherwise not offer certain interventions for various, non-moral reasons.32 In short, in many jurisdictions, there simply is no “graduation promise” that includes a commitment to performing every intervention that has been legalized within one’s practice area. While certainly there is a commitment to offer a wide array of interventions and passionately pursue the patient’s wellbeing, each CR case presents the question of whether that commitment includes this particular intervention in this particular context; concluding that a general patient-facing promise, without more, extends to a particular case is question-begging. The IT adherent remains convincing that we should be careful in carving out widespread medical practices or behaviors from the scope of the healthcare professional’s oath, which we want to be patient- and public-serving above all else. But the precise content of the professional’s commitment—and it is this precision that matters in CR cases—is an unavoidably context-dependent inquiry. Comparisons like Emanuel and Stahl’s, which claims that the obstetrician’s commitment to providing abortions is as self-evident as the hematologist’s commitment to providing blood transfusions, are unsupportable without such an inquiry.33 To put the difference succinctly, only the non-provision of the latter would constitute sanctionable medical malpractice in most jurisdictions. The fidelity argument, while powerful in justifying restrictions on CR, cannot justify a context-independent ban on it. Prevention of Harm to Patients and the Public Even if I am right that health care professionals across jurisdictions cannot be understood to take a sufficiently determinate oath to preclude CR, some IT adherents will reply that this misses the heart of the matter. Since CR is a refusal to offer a service, each instance of it directly causes or risks harms to individual patients and public interests. If we can make any cross-jurisdictional claims about healthcare at all, it is that healthcare practices are unequivocally about the patient’s well-being. Harms to patients include both physical harms, insofar as the unavailability of timely care puts their physical health at risk, and violations of their sense of worth and autonomy, insofar as CR claims can shame them for seeking desired treatment, foreclose treatment options that the law has made available to others, or cause them to incur additional costs in seeking alternate providers. Zolf (2019) cites two particularly troubling examples of this: (1) The case of Cheppudira Gopalkrishna, whose delayed request for physician-assisted dying (PAD) resulted in his continued, bedridden suffering of ALS for months, and (2) The prevalence of physicians who object to abortion in Italy, which results in approximately 20,000 abortions being performed there annually. While Zolf does not mention this sort of case himself, the possibility of CR in response to terminating ectopic pregnancies, which can endanger the mother’s life, is even more troubling.34 As a different kind of example, Ancell and Sinnott-Armstrong (who ultimately defend the compromise position) cite a case in which a same-sex couple is denied access to intrauterine insemination (IUI), causing them not just shame and indignity, but extraordinary expense in seeking alternative sources of treatment. What compounds the unacceptability of these harms, according to IT, is that they are cognizable in terms of medicine’s aims, whereas the countervailing values supporting CR are not. A patient’s prolonged suffering (in Gopalkrishna’s case), avoidable suffering (in the unsafe, illegal abortion cases), or bearing additional expenses that other patients do not (in the IUI case) are all clearly explicable in terms of B&C’s healthcare principles or bedrock democratic values (like non-discrimination). By contrast, the CR claims that brought those harms about are justifiable only through religious or moral views that are fundamentally distinct from medical judgments; they depend on conceptions of spiritual harm or a particular metaphysical view of fetal personhood, for example. Private, idiosyncratic worldviews cannot be wielded, especially by persons in such an authoritative position, to cause harms that are cognizable from a public, clinical perspective. In fact, allowing such a result violates the very publicity conditions offered in section “The Publicity Conditions”.35 Avery Kolers (2014) offers another compelling version of the private and public harms argument (hereafter, simply the “harm” argument), as he makes a concerted effort to proceed from shared political values. There are features of healthcare systems in most liberal democracies that all reasonable citizens recognize as morally important, including their ability to foster the autonomy of vulnerable persons (i.e., patients) and stably pursue other aims that society has deemed important. These features have not come about, nor will they remain, easily. Instead, they depend on “multiple autonomous agents” (i.e., healthcare professionals) coming together “under a system of rules to achieve morally important results that none of them could achieve by flouting those rules or by acting on their own or in a different context” (Kolers, 2014, 4). Thus, wherever a healthcare professional declines to participate in an intervention ordered by her superior, or where she elects to create her own, limited package of medical interventions that she offers—and does not offer interventions offered by her colleagues—this detracts from the high degree of cooperation a stable healthcare system requires. If such behavior where universalized, of course, this would result in disorder (i.e., an unpredictable healthcare system offering varied services) and a failure to foster vulnerable patients’ autonomy. To avoid such an outcome, individuals working in healthcare “cannot be unaccountable, and cannot be allowed to subvert the organization’s agenda or throw it into disarray” (Kolers, 2014,  4). Thus, to avoid these risks and serve the public interest, we rightly expect healthcare professionals to not just grudgingly submit to its practices, but to identify with the moral aims of the profession. Those who would seek to invoke CR, by contrast, live in a constant state of tension with the profession and constitute an ever-present risk of instability. The various iterations of the harm argument make a convincing case for certain restrictions or preconditions for CR claims, many of which have been discussed elsewhere. What is important for our purposes is that such restrictions can be justified democratically, or rather, through the reasoning process I outline in sections “The Publicity Conditions”–“Beyond the Publicity Conditions: A Constructive Approach to Deriving Healthcare Norms”. First, there should typically be no right to CR when it comes to emergent care and certain other, time-sensitive interventions. To hold otherwise likely violates all four publicity conditions outlined in section “The Publicity Conditions”, but especially the fourth, which requires a balancing of the parties’ interests where feasible (see, supra, p. 7). This would accord no weight to patients’ interests at all, as it is markedly different from circumstances of CR in which patients have the opportunity to seek out an alternate provider at little or no cost (and which therefore allow for meaningful balancing of the affected parties’ interests). This consequence would be difficult, if not impossible, to square with what the IT proponent is correct in describing as a public-serving social institution. Second, for similar reasons, the publicity conditions would seem to require objecting professionals to provide a degree of advance notice to patients, and perhaps even a duty to refer patients to nearby providers who do provide that intervention. While the literature contains entire debates about these options that I do not have space to address here, the publicity conditions would seem to require that affronts to patient time, energy, and expense are minimized. This might also include a governmental entity or healthcare employer bearing the cost of a patients’ having to seek alternative providers after being referred elsewhere. Third, any given practice area should be understood to have a substantive view of “good medicine,” complete with distinctions between core interventions under its purview that are obligatory for its physicians to provide versus those that are merely permissible or penumbral to the practice (and about which there is intentional room for a difference of opinion among colleagues). The trouble with IT, as I will further explain momentarily, is not in supposing there is an objective conception of good medicine within practice areas or jurisdictions; as the third publicity condition (comportment with shard history) suggests, resolution of any public issue needs such starting points. The trouble consists in IT accounts’ failures to derive their conception of good medicine from socially locatable sources or take heed of distinctions between permitted or obligatory conduct within practice areas. While these are not the only conceivable limitations that an inquiry along the lines of sections “The Publicity Conditions”–“Beyond the Publicity Conditions: A Constructive Approach to Deriving Healthcare Norms” might produce, they are three limitations that would seem to cross jurisdictional and other contextual boundaries in all liberal democratic healthcare systems. Notwithstanding their successful justification of limitations on CR like these, no iteration of the harm argument warrants a blanket (or near-blanket) ban on CR claims. The first reason for this is that, as Maclure and Dumont (2017) have argued, no IT account has ruled out the alternative of managing or limiting CR claims instead. Supposing that there are several ways of managing CR claims without causing patients significant harm, the fourth publicity condition (balancing parties’ interests) recommends the option that accords some weight to CR claimants’ interests. This is especially true if, as I argue below, CR claimants’ interests overlap with public interests. Moreover, while physicians play an important role in ensuring that patients have access to any care to which they are legally entitled, this burden is also shared with the government and other entities. The burden might instead fall on professional associations, state health authorities, or large healthcare employers to ensure that various regions are adequately staffed with professionals who are willing to provide the full range of services. Indeed, this is already the basis on which Certificates of Need are issued in the United States, whereby state or federal agencies ensure that new healthcare facilities meet the healthcare needs of the surrounding community.36 The trick, for democratic purposes (and pursuant to the first and fourth publicity conditions), is picking the burden-sharing route in patient access that is least coercive to all involved, a challenge that IT accounts typically show little interest in meeting. Proponents of the harm argument might reply that there is simply no good public reason to devote time and resources to managing CR claims, since they are only justifiable by private values that are immaterial to clinical decisions. This brings us to the second reason that the harm argument fails to justify IT: their mistaken assumption that worldview-neutral, clinical judgments are neatly separable from the parochial, moral ones that motivate CR. As both Cowley and Sulmasy have similarly argued, many CR claims are not simply motivated by extra-clinical or self-serving judgments; rather, they are motivated by the objecting professional’s clinical judgments, or more broadly, her understanding of the appropriate ends of medicine given her obligations to patients and the public. Admittedly the range of permissible interpretations of one’s clinical obligations must be constrained within each specialty, but the outer limits of that range—pursuant to the first and second publicity conditions—cannot be determined by the professional’s religious status. Even where such interpretations are religiously motivated, the most common types of CR claims (such as those involving abortion or PAD) are explicable in secular moral terms. The burden is on the IT proponent to explain why a particular CR claimant’s interpretation is outside this permissible range (from the relevant specialty’s perspective), which would require some recourse to the kind of procedure I have proposed in sections “The Publicity Conditions”–“Beyond the Publicity Conditions: A Constructive Approach to Deriving Healthcare Norms”. In fact, it is commonplace for clinical judgments to include both technical and moral components in other contexts. In some circumstances, there is intentionally no “official” or “preferred” interpretation of what one’s clinical duties require: there is intentional room for discretionary judgment. When it comes to judgments of medical futility, for example, critical care physicians’ judgments vary, even where they agree on the purely technical facts of the matter. Judgments about whether a particular procedure is worth performing, or who should receive priority of care, also appear in organ transplant committees’ decisions globally and in the United Kingdom’s judgments about which care ought to be publicly funded. What is most telling of all, however, is that IT proponents themselves defend the propriety of moral judgments in ways that closely resemble CR. Savulescu and Schuklenk (2017), for example, argue that it would be acceptable for a physician to attempt to talk her patient out of IVF—and instead opt for adoption—on the grounds that the latter better comports with distributive justice. Since their objection to CR is, at least in part, grounded in their claim that professionals should not interfere with or coerce patients’ informed healthcare decisions, this is a surprising recommendation. It is unclear why, given their arguments in favor of IT, it is the province of healthcare professionals to make judgments about which interventions comport with their private conception of distributive justice. My suspicion is that this constitutes their implicit recognition of the place of moral perspectives in clinical judgment: they simply object to those with a religious source. Portions of Kolers’s argument arguably survive this initial reply, as his account does not as assume the healthcare profession has a morally neutral view of harm, but instead grounds his account in the shared, political value of maintaining a stable, predictable, and publicly beneficial healthcare system. But there are value-judgments that even his iteration of the harm argument takes for granted. First, he seems to assume that a professional’s “identifying” with her profession’s “moral aims” necessarily means that she offers the maximal range of legalized interventions. As I have suggested, however, the moral aims of the profession often intentionally include the opportunity for continued disagreement over some interventions, particularly those that are controversial or novel. Moreover, it is not clear that we want a class of professionals who identify with all the practices common to their profession at a given time, for certainly we want motivated, conscientious healthcare professionals (rather than indifferent, conformist, or enterprising ones). It would be a surprising coincidence if a class of conscientious professionals did not produce vociferous disagreement on certain important matters at any given time. Second, while a stable and predictable healthcare system undoubtedly qualifies as one publicly intelligible value, it is far from the only one. It might be the case that some degree or risk of instability or non-uniformity in professional conduct is permissible on the grounds that it promotes other values, or even that it promotes a more enduring stability than the sort that aims at professional uniformity. Third and finally, it seems that stability, which on Kolers’s account is attained through uniformity in practices, is instrumentally, not intrinsically, valuable. In other words, he takes integrity (and stability) to be valuable because of the goods that such a healthcare system affords us. If there were an effective system for managing CR claims, however, then the instrumental value of stability would no longer constitute an objection to CR.37 Moral Consensus Within Healthcare Professions A third argument worth examining is better understood as a reply to a value that is invoked in defense of CR: religious and moral diversity among healthcare professionals. The claim that diversity in our social institutions is valuable has been mostly implicit in much of my discussion, though section “The Publicity Conditions” offered some general reasons supporting it (see Kolers, 2014, 4, 19–20). Against this claim, IT proponents have argued that there are affirmative reasons for circumscribing the degree of moral diversity within healthcare professions. We have already encountered an instance of this in one strand of Kolers’s argument. On his account, it is desirable to have sufficient moral uniformity among healthcare professionals for each of them to “identify” with the profession’s moral aims. This is a powerful argument with a Rawlsian legacy. Stephen Macedo (1998) has similarly argued, regarding religiously devout judicial appointees, that their public-serving vocation is a special one that requires some sacrifice of their personal aims in pursuit of distinctly civil interests. Thus, it is inevitable (and even desirable) that we will progressively find a particular range of worldviews within the judiciary and healthcare, especially one that might not reflect that of the citizenry. Other IT accounts are even less sanguine about the value of worldview diversity among healthcare professionals. In response to the claim that moral and religious diversity is in the public interest, Savulescu and Schuklenk (2017, 164) have argued that diversity should be reflected in the public sphere outside of healthcare, but not in the conduct of healthcare professionals. While they acknowledge that matters of controversy should remain open to public disagreement from those of diverse perspectives, these disagreements should be aired and resolved legislatively. All that should happen within healthcare, they claim, is the provision of all the services to which patients are “legally entitled” (Savulescu and Shuklenk, 2017, 164). Each of these arguments undervalue professional diversity because they undervalue the influence that professionals have on issues of public concern and the ways that professional diversity serves the public interest. Patients of various worldviews may benefit immensely from being able to seek out providers whose values resemble theirs, whether that is because such a provider is better equipped to foster their autonomy by helping them make important medical decisions (given their moral or spiritual priorities) or because the patient otherwise benefits from certain types of communication that only likeminded professionals might offer. Healthcare professionals are highly influential when it comes to initiating and advocating for legislative change, determining the priorities and direction of healthcare funding, leading and developing new areas for medical research, blowing the whistle on unethical practices within healthcare, and providing insight on the weight and urgency of health needs faced by the populations they serve. They also form part of institutional ethics committees, regulatory bodies, and professional associations. More broadly, the culture of healthcare itself is, and will always be, more than the regulatory and legislative measures that bind it: the moral and epistemic milieu of the field is largely shaped by those within it. For similar reasons, Rawls’s later work recognizes the vital role that the morally and religiously diverse “background” institutions of society play in forming the overlapping public consensus of diverse views that legitimize our system of government. Thus, not only do individuals of various background have an interest joining the ranks of the healthcare profession (an interest which is protected by the political value of equal opportunity), but various population groups’ interests—including the vulnerable patients themselves—are affected by the composition of these professions. The morally and religiously diverse public is also interested in healthcare having certain priorities and avoiding harmful practices, interests which are best served through a diverse professional class. The many limits of the legislative process mean that these matters are, in large part, entrusted to those who join the ranks of these professions. To suggest the relegation of moral and religious diversity to legislative conversations outside healthcare is to propose an impotent, even illusory, sort of diversity. It is also to render the profession dangerously uniform, as those who dissent from now-permissible practices are permanently foreclosed from entering it and serving as, among other things, a humbling check on the sometimes-unwarranted confidence of professional majority opinion. Both Sulmasy and Weinstock (2014) have, for this reason, argued that even those citizens with no qualms about current medical practices benefit from a diverse perspectival presence. What ultimately causes IT’s misappraisal of diversity’s value, as I suggested earlier, is their failure to take adequate stock of the plural values that are pertinent to deriving healthcare norms. When the professional objects to a practice, it is often far from a selfish, myopic concern for one’s own salvation, but is instead a judgment about precisely what is good for the patient, the public, or the profession’s future. Conclusion I have argued that many IT accounts, though they are well-motivated and persuasive in setting important limits on CR, fail to adequately ground their conclusions in their democratic milieus. The contrary position, Conscience Absolutism, even more transparently ignores various norms at the heart of healthcare professionalism. The CR debate should instead seek a framework for deriving healthcare obligations that is fair and sensitive to sociopolitical context. I have given reasons to believe that such analyses will tend to favor some version of the compromise approach, since its many iterations eschew generalized and context-independent answers to CR. This is not, as some IT accounts suggest, tantamount to moral relativism; it is grounded in liberal democratic values that recognize both the importance and limits of enshrining our own judgments about good medicine in law and public policy. In response to the CR question, we ought to be able to ground our position in one or several healthcare norms, and in response to the question of where those norms come from and how much they weigh, we ought to be able to give yet another answer, one that is grounded in something other than a bare expression of our own moral or spiritual worldview. Our answer, to invoke Dworkin’s chain novel analogy once more, should be in the spirit of joint authorship. Such an account of medical obligations is just as pressing in conversations tangential to CR, especially in an era that increasingly emphasizes social consciousness and public responsibility of healthcare professionals, especially in accounting for social determinants of health in their professional lives. Recently, the COVID-19 pandemic has highlighted the sacrifices and risks we ask of health professionals, as well as the extent to which their expertise might inform limits on citizens’ behaviors well outside the traditional domain of healthcare. If our moral expectations of healthcare professionals continually, even if gradually, shift under our feet, then we must agree on a means of determining what they are at any given moment, especially if such expectations are to be enshrined in law and public policy. I have also emphasized, in considering the myriad technological possibilities on the horizon, that healthcare professionals play an important role in determining the priorities and aims of our health institutions. Our history is already rife with examples of medical services whose widespread use or funding prioritization led to various forms of public harm, such as the opioid crisis in the United States. Now more than ever, medical research and practice might frequently find itself at numerous crossroads, in which various prospective medical practices or research directions might irreversibly shape our public health institutions for better or worse, and even alter what “better” or “worse” might mean in each case. While important decisions along these lines take place among the voting public, IT and Ab arguments often miss sight of two important considerations: (1) The public interest in intentionally leaving room for moral diversity and discretionary judgment within healthcare, and (2) The extent to which these decisions are unavoidably influenced by the healthcare community itself. Thus, for reasons of caution and democracy, we would do well to both construct our understanding of medical obligations carefully and ensure that we are on broadly justifiable ground wherever we restrict the prospect of dissent (via conscientious refusal) within our healthcare communities. 1 For a full description and survey of arguments for Ab and IT, see Wicclair (2011, Chs. 1–2). 2 For examples of these measures, see Wicclair (2011) and Lynch (2008). 3 See, e.g., Schuklenk and Smalling (2017) and Kolers (2014). 4 See, e.g., Stahl and Emanuel (2017). 5 See Kolers (2014). 6 This would also rule out what Wicclair has called “conscience absolutism,” which is the view that provider conscience always trumps patient demand. I have omitted much discussion of this approach because it is virtually unrepresented in the academic or professional literature. Influential views defending CR recognize important, contextual limits on its use. 7 For a detailed discussion and defense of the compromise position, see Wicclair (2011). 8 There are some recent exceptions to this, particularly Robert Card & Doug McConnell’s applications of Rawlsian public reason [see Card & McConnell (2019) and McConnell (2019)] and Nir Ben-Moshe’s application of Adam Smith’s impartial observer framework to CR [see Ben-Moshe (2019)]. A recent issue of the HEC Forum (Vol. 33, No. 3) also includes some contributions that are promising in this regard. Some IT accounts [such as Zolf (2019) and Savulescu and Schuklenk (2017)] have responded by invoking political values, but only in a cursory manner. 9 As noted above, the reason for focusing on these IT accounts is not because this sort of methodological mistake is unique to IT positions: it is in part because the contrary position, Ab, is simply not well represented or articulated in the literature. Nonetheless, section “Beyond the Publicity Conditions: A Constructive Approach to Deriving Healthcare Norms” will offer some insights into why Ab would not pass muster under the methodological framework I propose. 10 See, e.g., Savulescu and Schuklenk (2017 163) (“It is clear that the scope of professional practice. is ultimately determined by society, and that it is bound to evolve over time.”); Sulmasy (2019, 16) [“In dialogue with society, professions (in liberal democracies) establish the goals of their practice and the ethics of their practice, recognizing the discretionary space necessary for individual practitioners.”]. 11 See, e.g., Zolf (2019, 150). 12 See, e.g., Sulmasy (2019) and Powell (2019). 13 I intend for this criticism to apply equally to those accounts that mention certain political or jurisprudential concepts in a conclusory manner, rather than employing them through the sort of detailed analysis that such traditions warrant. 14 By contrast, pursuant to what he called the “fact of oppression,” Rawls surmised that ideological uniformity among citizens was a sign of democratic failure. 15 See Rawls (2005, 8) for a discussion of “considered convictions” as the foundation for any debates about public policy in liberal democracies. For Rawls, it is beyond the ken of political philosophy to call considered convictions into question or ground them in a deeper, metaphysical foundation. 16 See, e.g., Muldoon (2016) for both empirical and philosophical arguments about the ubiquitous benefits of perspectival diversity. 17 For a clear violation of this condition, see Fiala and Arthur (2017), wherein the resolution to the CR debate depends on the decline of organized religion. As Habermas has suggested in a different context, this would also seem to be a risk inherent in any IT account which insists on religious or moral convictions being best understood as oddities or peccadillos. 18 For contemporary discussions of different conceptions of public reason, see Vallier (2014) and Eberle (2002). 19 It is worth emphasizing here that reasonableness is only meant in the moral and not the epistemic sense. As such, reasonableness is not an inquiry into the rationality or evidentiary basis of the worldviews in question. 20 This is similar to the attitude that Rawls (2005) took, in Political Liberalism, to his model of “justice as fairness,” which he considered one of several of a “family” of conceptions of justice that might satisfy the requirements laid out in Political Liberalism. 21 As one example among many, it’s possible that a professional organization’s position statement on an issue or a provision of one of its codes of ethics are products of cabalistic attempts to flout public interests and popular opinion. 22 For a similar observation in the judicial context, see Dworkin (1986, 255). 23 There is nothing inherently wrong with attempts to describe an ideal medical practice and argue for departures from anything resembling our existing paradigm. But this is generally not what the IT accounts I am criticizing take themselves to do: even if part of their account is aspirational, much of it purports to describe medical obligations as liberal democratic society currently, as a matter of social fact, understands them. 24 I am grateful to an anonymous reviewer for putting forward this important challenge and providing the opportunity to answer it. 25 For a similar argument about matters of religious pluralism and public policy generally, see Tebbe (2017 42). 26 Moreover, insofar as the citizenry is sufficiently emphatic that practitioners and professional associations are “behind the times” regarding a particular issue, the democratic lawmaking process offers a mechanism for fundamentally altering the normative environment. Thus, in the absence of a law to the contrary, there are reasons for according significant weight to some preponderance of professionals themselves. 27 Ben-Moshe (2021) has recently proposed that a diverse panel called the Uber Conscientious Objection in Medicine Committee (UCOM) be instituted for adjudicating CR claims. The framework this article defends might supply an appropriate reasoning process for such an adjudicative body’s deliberations about CR. 28 While disputes about which practices are core or essential are themselves difficult, it is more difficult to defend the position that there is no such distinction. As I suggested earlier, this requires contextual and fact-specific analyses in various cases. 29 Notable examples of this include the Tuskegee Syphilis Study, the Willowbrook School Hepatitis Study, and the Jewish Chronic Disease Hospital case, the aftermath of which heavily influenced the Belmont Report and contemporary research ethics. 30 See Savulescu and Schuklenk (2017, 165) for an example of this suggestion. Kolers (2014) analogously suggests that professionals’ uniformity in offering all legalized services in their specialty furthers the aims of medicine. 31 See Cowley (2016, 361) for a similar point. 32 As Ancell and Sinnott-Armstrong (2017), Powell (2019), and Sulmasy (2019) each observe, it is commonplace in some jurisdictions for professionals to define the scope of their own practice. 33 See Emanuel and Stahl (2017, 1381). 34 This example has become more realistic and pronounced in a post-Dobbs United States. 35 See Emanuel and Stahl (2017) for a similar argument. 36 Analogically, Lynch (2008) proposes a system in which licensing boards would ensure proportional representation of different worldviews within the various medical professions. 37 To be fair to Kolers, he might agree with much of this analysis, as he is importantly different in orientation than other IT approaches (and is arguably not a strict IT proponent at all): he finds allowing some CR permissible, but only as a license rather than a right. Nonetheless, engaging Kolers as an IT proponent is appropriate to the extent that I defend certain CR claims as a right (which seems to be a necessary extension of compromise approaches). Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References Ancell A Sinnott-Armstrong W How to allow conscientious objection in medicine while protecting patient rights Cambridge Quarterly of Healthcare Ethics 2017 26 1 120 131 10.1017/S0963180116000694 27934570 Beauchamp T Childress J The principles of biomedical ethics 2013 7 Oxford University Press Ben-Moshe N The truth behind conscientious objection in medicine Journal of Medical Ethics 2019 45 404 410 10.1136/medethics-2018-105332 31221763 Ben-Moshe N Conscientious objection in medicine: Making it public Hec Forum 2021 33 269 289 10.1007/s10730-020-09401-z 32221817 Card R McConnell D Public reason in justifications of conscientious objection in healthcare Bioethics 2019 33 5 625 632 10.1111/bioe.12573 30865301 Cowley C Defense of conscientious objection in medicine: A reply to Schuklenk & Savulescu Bioethics 2016 30 5 358 364 10.1111/bioe.12233 26659648 Dworkin R Law’s empire 1986 Belknap Press Eberle C Religious convictions in liberal politics 2002 Cambridge University Press Fiala, C. & Arthur, J. (2017). There is no defence for 'Conscientious objection' in reproductive healthcare. European Journal of Obstetrics & Gynecology and Reproductive Biology, 216, 254–258. Gaus G Justificatory liberalism 1996 Oxford University Press Giubilini A Objection to conscience: An argument against conscience exemptions in healthcare Bioethics 2017 31 5 400 408 10.1111/bioe.12333 28008640 Kolers A Am I my profession’s keeper? Bioethics 2014 28 1 1 7 10.1111/bioe.12056 24117529 Lynch HF Conflicts of conscience in healthcare: An institutional compromise 2008 MIT Press Macedo, S. (1998). Transformative Constitutionalism and the Case of Religion: Defending the Moderate Hegemony of Liberalism. Political Theory, 26(1), 56–80. Maclure J Dumont I Selling conscience short: A response to Schuklenk and Smalling on conscientious objection by medical professionals Journal of Medical Ethics 2017 43 4 241 248 10.1136/medethics-2016-103903 27681301 McConnell D Conscientious objection in healthcare: How much discretionary space best supports good medicine? Bioethics 2019 33 1 154 161 10.1111/bioe.12477 30014476 Muldoon R Social contract theory for a diverse world: Beyond tolerance 2016 Routledge Powell K Reasonable accommodation of conscientious objection in health care is morally and legally required Perspectives in Biology and Medicine 2019 62 3 489 502 10.1353/pbm.2019.0028 31495793 Quong J The rights of unreasonable citizens Journal of Political Philosophy 2004 12 3 314 335 10.1111/j.1467-9760.2004.00202.x Rawls J Political liberalism 2005 expanded Columbia University Press Savulescu J Schuklenk U Doctors have no right to refuse medical assistance in dying, abortion or contraception Bioethics 2017 31 3 162 70 10.1111/bioe.12288 27716989 Schuklenk U Smalling R Why medical professionals have no moral claim to conscientious objection accommodation in liberal democracies Journal of Medical Ethics 2017 43 4 234 10.1136/medethics-2016-103560 27106748 Stahl R Emanuel E Physicians, not conscripts—Conscientious objection in health care New England Journal of Medicine 2017 376 14 1380 1385 10.1056/NEJMsb1612472 28379789 Sulmasy D Conscience, tolerance, and pluralism in health care Theoretical Medicine in Bioethics 2019 40 507 521 10.1007/s11017-019-09509-5 Talisse R Democracy after liberalism 2009 Oxford University Press Tebbe N Religious freedom in an egalitarian age 2017 Harvard University Press Trigg R Conscientious objection and ‘effective referral’ Cambridge Quarterly of Healthcare Ethics 2017 26 1 32 43 10.1017/S0963180116000633 27934565 Vallier K Liberalism and public faith: Beyond separation 2014 Routledge Wicclair M Conscientious objection in healthcare 2011 Cambridge University Press Weinstock D Conscientious refusal and religious professional: Does religion make a difference? Bioethics 2014 28 1 8 15 10.1111/bioe.12059 24117664 Zolf B No conscientious objection without normative justification: Against conscientious objection in medicine Bioethics 2019 33 1 146 153 10.1111/bioe.12521 30256432

36520271

PMC9753870

NO-CC CODE

2022-12-16 23:26:35

no

HEC Forum. 2022 Dec 15;:1-31

utf-8

HEC Forum

10.1007/s10730-022-09502-x

oa_other

==== Front Educ Inf Technol (Dordr) Educ Inf Technol (Dordr) Education and Information Technologies 1360-2357 1573-7608 Springer US New York 11514 10.1007/s10639-022-11514-6 Article An improved accurate classification method for online education resources based on support vector machine (SVM): Algorithm and experiment http://orcid.org/0000-0002-3892-7261 Quan Zhi quanzhi@swufe.edu.cn 1 Pu Luoxi 2 1 grid.443347.3 0000 0004 1761 2353 Southwestern University of Finance and Economics, Chengdu, China 2 grid.444472.5 0000 0004 1756 3061 UCSI University, Kuala Lumpur, Malaysia 15 12 2022 115 22 8 2022 5 12 2022 © The Author(s), under exclusive licence to Springer Science+Business Media, LLC, part of Springer Nature 2022, Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. In the face of surging online education around the globe, it seems quite necessary and helpful for learners and teachers to have the plethora of online resources well sorted out beforehand. To some extent, the efficiency and accuracy of resource search and retrieval may determine the quality and influence of online education. In this research, based on the methodological framework of design science, the support vector machine (SVM) algorithm is chosen to optimise the design of an accurate resource classifier. The aim is to improve the unsatisfactory classification effect of traditional classification methods for online education resources, so that online learners can enjoy more accurate and convenient access to education resources they are seeking out of many more. For the purpose of performance evaluation, the proposed SVM-based classifier was compared with two other classification methods based on multiple neutral networks and deep learning respectively. Upon collection and pre-processing of online materials, the features of educational resources were extracted and output in the form of feature vectors. By calculating the similarity between the extracted feature vectors and the standard vectors of the set type, we obtained the classification results of online education resources for each of the three classifiers. It was found that, compared with those of the two traditional classification methods, the precision ratio and the recall ratio of the proposed classifier improved by 3.26% and 2.01% respectively. In the meantime, the proposed SVM-based classifier seems to more advantageous in performance balance with better F measurement. Keywords Support vector machine (SVM) Design science Online education resources Accurate classification ==== Body pmcIntroduction In the past decades, online education has become a global phenomenon by penetrating into virtually all countries in the world with the expansion of the Internet infrastructure (Chan et al., 2022; Dhawan, 2020; McCarty et al., 2006; Rye, 2014). This trend is particularly the case for China, with perhaps the largest population of learners on the Earth. According to a 2018 report, up to 144 million people had taken up online education as of June 2017 in China (Yang & Du, 2018); the number surged amazingly to over 300 million (including millions of teachers) in 2020 after the breakout of the COVID-19 pandemic, when online delivery was almost the only choice for both educators and students then (Li, 2020b). This world’s largest ICT-based teaching experiment and reform have been continuing in the post-epidemic era, offering a wealth of diversified education resources through the Internet (Li, 2020a). Online education, receiving unprecedented and ever-increasing attention from educators, students and the public, has become integral to the whole educational system. The abundant materials provided by all kinds of online education platforms bring both benefits and challenges to users. On the one hand, teachers and students are enabled to employ online courses to complete pedagogical processes. On the other hand, it tends to be rather time-consuming to retrieve the right materials that one is seeking out of a huge amount of online resources. To some extent, the efficiency of resource search and retrieval may determine the quality and influence of online education. Thus, enriched content entails enhanced classification methods for effectively and efficient delivery of online education, otherwise many learners may get lost and bored, and even daunted, when wading through massive data of online materials. In the practice of classifying online education resources, there seem to be problems of incomplete coverage of fields and unscientific categorisation, among others. In this research, an accurate classification method based on support vector machine (SVM) is proposed in order to improve the utilization of online education resources. In a properly built model for autonomous learner users, SVM has the potential to help enhance resources classification and allocation of online education resources. Delving into the optimisation of SVM is hence of significant educational and research value. A brief review of relevant literature is set out below, followed by the proposed algorithm and results of a comparison experiment to evaluate the effectiveness and efficiency of the SVM-based classification algorithm. Literature review Classification can be seen as a process to summarise the features and classification rules based on sample data sets of existing resources and to establish rules of discrimination, so that new resources can be categorised according to such established rules (see Diederich, 2008). To classify the ever-growing body of online education resources may be a task beyond human capacities. We may take China as an example: years ago, it was planned that by 2020, there would be over 3,000 national-level courses available online (Yang & Du, 2018); the amount of online courses offered by all levels of education providers may become even multi-fold of that now. In addition, each course may include multiple formats of materials: texts, images, audios, videos and so on, which exacerbates the complexity and difficulty of data search and retrieval. In this case, it seems a good solution to train machines to analyse such large and complex datasets (Shalev-Shwartz & Ben-David, 2014), and document classification has been a traditional task that machine learning can deal with satisfactorily (Mehryar et al., 2018). In history, international research on the classification methods of education resources had an early start. Methods based on word frequency statistics and factor analysis have been widely used in email, information retrieval, and so on (see Joachims, 2002). Other traditional methods of more complex resource classification may have to rely on multiple neural network integration and deep learning (Lam et al., 2012). However, some problems, such as incomplete coverage of fields and unscientific categorisation, can be identified in practice, which ultimately affects the quality of their application. Thus, the support vector machine (SVM) is adopted as the basis for the proposed classifier in this research, since this algorithm may help solve the extremum problem in the traditional methods (Hamel, 2009). SVM is an established method in natural language process (NLP). It is one of the most significant kernel-based methods of machine learning and one of the most popular supervised learning algorithms, widely used for classification and regression analysis (Steinwart & Christmann, 2008). A classic definition is set out as follows (Cristianini & Shawe-Taylor, 2000, p. 7):Support Vector Machines (SVM) are learning systems that use a hypothesis space of linear functions in a high dimensional feature space, trained with a learning algorithm from optimisation theory that implements a learning bias derived from statistical learning theory. This learning strategy introduced by Vapnik and co-workers is a principled and very powerful method that in the few years since its introduction has already outperformed most other systems in a wide variety of applications. Given the agreed-on outperformance of SVM over other methods, it is not surprising that SVM has been widely used in research of various fields, including but limited to financial analysis, medical analysis, biology, so on and so forth (Ma & Guo, 2014; Murty & Raghava, 2016; Suykens et al., 2015; Wang, 2005). Here is a basic explanation and illustration of the fundamental principle of SVM for classification, as shown in Fig. 1. The central task of SVM is to create a hyperplane between data sets to indicate which class an item probably belongs to. The challenge is to train the machine to understand structure from data and map with the right class label. For the best result, the hyperplane has the largest distance to the nearest training data points of any class. Thus, the classification based on the SVM algorithm can be considered as a hyperplane made up of multiple separate heterogeneous data samples, to which a solution can be worked out. Through the calculation of the maximum spacing of heterogeneous samples, the category of the target sample can be determined, and upon further processing the classification of all samples can be completed.Fig. 1 The principle of classification using the SVM algorithm The signs " + " and "-" in Fig. 1 represent distributed and negative data samples respectively. It can be seen in the graph that, H points to the centre of the adjacent edge between the two types of samples, serving as the isolation barrier for different types of samples; L1 and L2 are located on the verge of the two types of samples, and the distance between them shows the classification margin. During the process of solving the maximum value, if the dividing line H can completely separate positive and negative samples like a watershed, H is then optimal (Liu et al., 2019). Experimental results show that, if the classification samples are distributed in multi-dimensional space, on the assumption that there is hyperplane H-value completely separated with the most significant classification interval, and that there are two types of classification – ‘positive’ and ‘negative’, then the classification interval is the largest, which can also be used to predict other data of the same classification. Deng et al. (2013) provide more comprehensive and extended elaboration on the mathematic representation of SVM. Certainly SVM is not perfect from its inception; it has been embracing continuous improvement with researchers’ effort. Wiering and Schomaker (2015) point out the limitations of standard SVM: a shallow model with a single layer of support vector coefficients, and overreliance on inflexible kernel functions; instead, they propose a transition from the single-layer SVM to the multi-layer SVM with deep architectures. There have also been proposals to integrate SVM with other established algorithms (e.g. Stoean & Stoean, 2014). In a word, iterative progress can make SVM a refined tool for both research and practice. Although it may not be the most novel or popular one, there seems to be much room to optimise this method based on previous and existing endeavours. Methodology This research follows the methodological framework of design science research (DSR) or design research. DSR has been widely used in IT engineering for decades (Gregor, 2021; Hevner et al., 2004; Peppers et al., 2007), and it features intentional creation and development of artifacts that serve human purposes (Dresch et al., 2015), aiming to “change the state-of-the-world through the introduction of novel artifacts” (Vaishnavi & Kuechler, 2008, p. 18). Apart from tangible and concrete solutions to identified problems in human activities, DSR also seeks new understanding through dynamic interaction of artifacts and knowledge. This research aims to develop an optimised classifier with higher classification accuracy of online education resources and gain new knowledge on how to best classify massive resources for best delivery of online education. The ultimate goal of the proposed method is to categorise the disciplines, specialisation and chapters of such resources based on the content. However, different education resources rely on various storage formats, including texts, videos, web pages, images, etc. It is helpful to target online education resources in different multimedia formats and determine the specific types via content analysis and feature analysis. Prior to the design work, it is necessary to identify and set rules for online resources of different disciplines as the reference standards for classification. Automatic collection Appropriate methods of automatic collection and processing of online education resources are entailed. The first step is to collect the target online education resources on the Internet using web crawlers. Compared with traditional methods of data collection, web crawlers can start to search webpages on the Internet according to pre-set URLs and extract the links in the pages. Then new links can be retrieved, and education resources can be downloaded automatically. The initial collection of education resources involves data including texts, videos and webpages, and the main module of web crawlers supports webpage data download and data parsing. In the actual collection process, the action is to initiate the downloader module of web crawlers according to set parameters and read the first URL. According to the results, online education resources are searched on the Internet. Comparison with data in the local resource library is then conducted upon identification of the online resource location, so as to check if the online resource has already been stored in the local repository. If it is already included, there is no need to repeatedly download it, otherwise it will be downloaded and stored in the local repository. During the comparison of education resources, apart from the resource names, the size and update date/time should also be compared, to ensure that the collected resources are of the latest versions. Upon the completion of a round of collecting resource data, the corresponding URLs are stored in the list of assessed resources. The next step is to parse the newly downloaded data using the parser in the crawlers and extract the URL information. If the extracted URL is existing, it is deleted immediately, otherwise it is stored in the list of resources to access. Upon the completion of resource downloading following the above procedure, another URL is extracted and the above procedure is cycled. All the online education resources are collected from the Internet, and collected data are stored separately according to different storage types. Processing of collected resources It is helpful to clarify the processing of online resources in text, image and video formats. The purpose of processing text-format educational resources is to extract the target data and convert it to row format. The units of text data are words, phrases, paragraphs and so on, and text information exactly consists of such units of natural language. In the process of text feature representation, it is a major step to extract the noun phrases and proper nouns, like names of people and places in the texts. There might be a large number of characters irrelevant to the central ideas of the texts, such as numbers, links, punctuation marks, and stop words. In order to reduce the complexity of processing text data, it is helpful to have the semantic vocabulary set of highly conceptual texts as the textual feature set and replace original text with the textual features. To ensure the effective information content in the textual feature vector, pre-processing text data is an important procedure worthy of further research. The next step is to filter the stop words in the original texts. Stop words can be divided into two kinds: undistinguishable words and function words. Undistinguishable words refer to high frequency words in almost all types of texts. Function words include pronouns, participles, and so on. Upon establishment of the stop word list, the matched words will be deleted, while the unmatched will be retained in the keyword list. Preprocessing image-format education resources Pre-processing image-format educational resources takes two steps: the one is screening based on image quality, and the other is unified processing of image formats. Simple filtering comes first during the quality-based screening of images, where median filtering processor is employed. If the output image resolution is higher than 75%, it will be retained, otherwise the image will be removed. The unified processing of image formats can be divided into two aspects: the unification of image storage formats and image colours. It is set that the storage format of educational resources should be JPG and the colour space should be RGB. Pre-processing video-format education resources To guarantee the classification efficiency of educational resources in video formats, it is necessary to mine valuable information frame-by-frame in the video resources, which can be deemed as a video-format bag of words. Figure 2 illustrates the text mining process of video-format educational resources.Fig. 2 Flow chart of text mining of video education resources Mining of image information in video-format educational resources can be conducted in the same way. Then video-format educational resources can be pre-processed following the above methods for text-format and image-format educational resources. Extracting features online education resources This section addresses document frequency, information gain and word frequency. Firstly, document frequency refers to the number of documents in the collection that contain a term. The larger the number is, the more frequently the certain term appears in documents, and the more the feature words contribute to classification. This can be used as an important criterion for classification. After word filtering, the dimensions of text vector are reduced, with little impact on the classification accuracy. The extraction of feature words can help reduce the dimensions of vector space and amount of calculation, which can indirectly improve the efficiency and accuracy of text classification. Secondly, information gain of textual resources refers to the changed amount of entropy with the creation of texts, which is an important part of text classification. The features of information gain extracted from online educational texts can be expressed as:1 μIGw=PW∑i=1mPCi|WlgPCi|W-∑i=1mPCilgPCi+PW¯∑i=1mPCi|W¯ In Eq. (1), Ci and W stand for class variables and features respectively. The variables P(Ci|W) and PW¯ are the probabilities of the text falling into the category Ci under two conditions: including or excluding the feature W; PCi|W¯ is the conditional probability of the text falling into the category C when W is not included. Thirdly, word frequency refers to the frequency of target words in textual educational resources, which can be calculated as follows:2 μWt,d→=tft,d→×lgNnt+0.01∑t∈d→tft,d→×lgNnt+0.012 In Eq. (2), μWt,d→ refers to the weight of the word t in the text d→; the variable tft,d→ is the word frequency; N and nt stand for the total numbers of training texts and those found to include the word t. Building an accurate classifier of resources using the SVM algorithm Based on the fundamental SVM illustration in Fig. 1, an optimised accurate classifier can be built with the following steps. We assume that, as per the set of training texts of online educational resources, the initial input is (xn, yn), and then the division of the hyperplane H can be shown using the linear equation in Eq. (3) in the two-dimensional space.3 uTx+b=0 In Eq. (3), u and b are the normal vector and offset value of the linear equation respectively. The samples on L1 and L2 in Fig. 1 are defined as the support vector, and 2‖u‖ represents the classification margin. Then the maximization problem of online educational resource classification can be converted to a corresponding dual problem: with constraints set, the task is to find out the maximum value of the dual function.4 ∑i=1nαiyi=0Lα=∑i=1nαi-12∑i=1nαiαjyiyjxiTxj The vector α in Eq. (4) is the language multiplier corresponding to the samples in the training data. With the solution of the vector α, the specific values of the parameters of u and b in the optimal hyperplane function. The kernel function of null SVM is K(xi, xj), based on which the function of an accurate classifier for non-linear resources can be developed as:5 fx=∑i=1nαiyiKxi,xj+b Lastly, the iterative training procedure of the proposed SVM-based optimised classifier is shown in Fig. 3.Fig. 3 The iterative training procedure of the proposed resource classifier Implementing classification of online educational resources This step is to have the automatically collected and pre-processed online educational resources as the entries and load them chronologically into the SVM-based classifier. According to the features of the loaded resources, with the classifier serving as the running environment, the degree of similarity between the feature vector extracted and the features of the target categories can be calculated following this equation:6 SimD0,Di=∑k=1pμki×μkj In Eq. (6), μki and μkj are the standard feature vector of a certain category D0 and the comprehensive feature vector extracted from loaded samples. If we compare the results from Eq. (6) with the pre-set similarity threshold, the category with calculated value above the threshold can be used to label the specific educational resource. If there are more than one category that can meet the criterion, the one with the highest similarity will be used as the final result of classification (Beyene et al., 2020). Results from a comparison experiment A comparison experiment was designed to test the classification accuracy of the proposed SVM-based accurate classifier of online educational resources. The counterparts in the experiment were the resource classification method based on fusion of multiple neural networks and the method based on deep learning (Lam et al., 2012). In order to reduce the impact of independent variables on the results, the online education platform and raw samples of resources were identical. Configuration of the online education platform The experiment used the online education platform of a university located in the southwestern region of China as the environment. The platform consists of several client sides for students and teachers, and one server and one database. All the resources of the online education platform are stored in the database. In the experiment, the three classification methods (the SVM-based one, and the ones based on fusion of multiple neural networks and deep learning) were translated into codes and embedded into the online education platform. Figure 4 illustrates the configuration of the classification methods in the experimental environment.Fig. 4 Configuration of the classification methods of online education resources The two experimental comparison methods can also be imported and configured following the same way. In order to guarantee the independence of the three classification methods, parallel running was selected to implement the calling and switching among the different methods. The samples of online educational resources used in this experiment were taken from two sources: the database of the education platform of the above-mentioned university, and the database of the University of California, Irvine. Several experiments were conducted on multiple data sets from the above database, to find more accurate results of classification. Upon statistical processing, the size of the sample online educational resources in this experiment amounted to 254.65 GB, all for the discipline of mathematics. The types of resources included texts, tables, images, videos, audios and so on. After that, three indicators: precision ratio, recall ratio and F measurement, were set as the indicators to evaluate classification results. Recall ratio refers to the proportion of correctly categorised documents after classification, while precision ratio refers to the ratio between the number of correctly categorised documents and the number of documents expected in that category. The results of precision ratio and recall ratio are shown as P and R. The quantitative results of the two indicators can be expressed as:7 P=TPTP+FP×100%R=TPTP+FN×100% In Eq. (7), TP refers to the amount of correctly categorised resources, while FP is the amount of incorrectly categorised resources, and FN refers to the amount of unclassified resources that fall into the category. The last indicator F measurement is used to measure the balance between precision ratio and recall ratio, which can be expressed as:8 FβP,R=β2+1PRβ2P+R In Eq. (8), β is the adjustment parameter. In general, the higher the F measurement is, the more balanced the precision ratio and recall ratio are. Result analysis of the comparison experiment As designed, the different classification methods were imported into the experiment environment. Upon debugging it was ensured that the methods could run in the experiment environment. Classification results of the proposed SVM-based classifier, together with those of the other two methods, can be worked out following the same procedure. Then we compared the results with the quantitative results of the set indicators, as shown in Table 1.Table 1 Testing results of classification performance of online education resources Times of experiments 1 2 3 4 5 Classifier based on fusion of multiple neural networks TP/GB 46.27 43.44 45.62 42.77 44.01 FP/GB 2.81 2.79 2.94 2.85 2.76 FN/GB 1.85 1.67 1.57 1.72 1.74 Classifier based on deep learning TP/GB 49.35 48.62 47.17 48.04 49.22 FP/GB 2.27 2.35 2.19 2.31 2.28 FN/GB 1.44 1.29 1.35 1.41 1.33 Proposed SVM-based classifier TP/GB 51.86 61.39 55.82 54.07 62.08 FP/GB 1.22 1.05 1.16 1.21 1.07 FN/GB 0.59 0.67 0.82 0.64 0.55 Putting the data in Table 1 into Eq. (7), we can work out the average precision ratio and recall ratio of the three classification methods, respectively as: 94.01% and 96.29%; 95.51% and 97.26%; 98.02% and 98.86%. It can be seen that the proposed SVM-based classification method achieved some improvement in both precision and recall ratios. If we input the results of precision ratio and recall ratio into Eq. (8) and determine the parameter β as 1, the quantitative results of F measurement on different data sets can be worked out, as shown in Fig. 5.Fig. 5 Testing results of F measurement As can be seen from Fig. 5, the F measurement result of the proposed SVM-based accurate classification method for online educational resources is always higher in all data sets than those of the other two methods. In other words, the classification performance of the proposed classifier tends to be more balanced. It implies that, the proposed accurate classifier of online educational resources may achieve the least exclusion of possible samples to the utmost degree of classification accuracy. Conclusions In a nutshell, the SVM-based classifier designed in this research can achieve slight improvement in both precision ratio and recall ratio of resource classification, when compared with the traditional methods. The classifier can reconcile the two seemingly conflicting indicators to a large extent and produce more balanced output with more accurate and inclusive results of classification. That is, by using this SVM-based accurate classifier, users may be provided with some more resources put into the correct categories at a time. It is believed that easier, more decentralised and engaging access to online resources is the trend of ICT-enhanced education (Fox, 2011). As the designed classifier can enhance the classification results of online education resources, it will thus indirectly improve the retrieval efficiency and usability of such resources. The limitations of this research may invite further effort on this topic. Firstly, all the samples are of mathematics only. If the algorithm is applied to other disciplines distinctively different from mathematics, e.g., visual arts, literature, the testing results may differ. In the meantime, online materials for cross-disciplinary subjects, such as behavioural finance and computational linguistics, are not considered in this experiment. Secondly, no more major methods apart from SVM are involved in this research, while the synergy of hybrid algorithms may help further improve classification results, e.g., combining SVM with deep learning techniques. All in all, optimal results of classification entail ongoing and iterative optimisation of accurate classifiers, including this SVM-based one. To achieve this goal, programming integrating multiple methods can be tested on materials from multiple disciplines to identify the best solution and best practice. Data availability The data supporting the findings of this study are available upon reasonable request. Declarations Conflicts of interest/Competing interests No potential conflict of interest was reported by the authors. Publisher's note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References Beyene, W. M., Mekonnen, A. T., & Giannoumis, G. A. (2020). Inclusion, access, and accessibility of educational resources in higher education institutions: Exploring the Ethiopian context. International Journal of Inclusive Education, 1–17. 10.1080/13603116.2020.1817580 Chan RY Bista K Allen RM Online teaching and learning in higher education during COVID-19: International perspectives and experiences 2022 Routledge Cristianini N Shawe-Taylor J An introduction to support vector machines and other kernel-based learning methods 2000 Cambridge University Press Deng N Tian Y Zhang C Support vector machines: Optimization based theory, algorithms, and extensions 2013 CRC Press Dhawan S Online learning: A panacea in the time of COVID-19 crisis Journal of Educational Technology Systems 2020 49 1 5 22 10.1177/0047239520934018 Diederich J Rule extraction from support vector machines 2008 Springer Dresch A Lacerda DP Antunes JAV Jr Design science research: A method for science and technology advancement 2015 Springer Fox, R. (2011). Technological practice and change in education. In R. Kwan, C. McNaught, P. Tsang, F. L. Wang, & K. C. Li (Eds.), Enhancing learning through technology - Education unplugged: Mobile technologies and Web 2.0 (International Conference, ICT 2011 Proceedings) (pp. 1–7). Springer. Gregor S Aier S Rohner P Schelp J Reflections on the practice of design science in information systems Engineering the transformation of the enterprise: A design science research perspective 2021 Springer 101 113 Hamel LH Knowledge discovery with support vector machines 2009 Wiley Hevner AR March ST Park J Ram S Design science in information systems research MIS Quarterly 2004 28 1 75 105 10.2307/25148625 Joachims T Learning to classify text using support vector machines 2002 Springer Lam HK Ling SH Nguyen HT Computational intelligence and its applications evolutionary computation, fuzzy logic, neural network and support vector machine techniques 2012 Imperial College Press Li C Online education in the post-epidemic Era International Journal of Social Science and Education Research 2020 3 10 332 335 10.6918/IJOSSER.202010_3(10).0052 Li, Y. (2020b). Booming online education. China Today, pp. 56–59. Retrieved from http://www.chinatoday.com.cn/ctenglish/2018/sl/2020b06/t2020b0605_800208979.html Liu M Liu X Li S Research on the resource planning procedure of education cloud platforms: Take enterprises of digital education resources as an example [(in Chinese) Jiao Yu Yun Ping Tai Zi Yuan Gui Hua Liu Cheng Yan Jiu: Yi Shu Zi Jiao Yu Qi Ye Wei Li] E-Education Research 2019 40 11 78 85 Ma Y Guo G Support vector machines applications 2014 Springer McCarty S Ibrahim B Sedunov B Sharma R Weiss J Nolan J Hunsinger J Trifonas P Global online education The international handbook of virtual learning environments 2006 Springer 723 787 Mehryar M Rostamizadeh A Talwalkar A Foundations of machine learning 2018 2 The MIT Press Murty MN Raghava R Support vector machines and perceptrons: Learning, optimization, classification, and application to social networks 2016 Springer Peppers K Tuunanen T Rothenberger MA Chatterjee S A design science research methodology for information systems research Journal of Management Information Systems 2007 24 3 45 77 10.2753/MIS0742-1222240302 Rye SA The educational space of global online higher education Geoforum 2014 51 6 14 10.1016/j.geoforum.2013.09.011 Shalev-Shwartz S Ben-David S Understanding machine learning: From theory to algorithms 2014 Cambridge University Press Steinwart I Christmann A Support vector machines 2008 Springer Stoean C Stoean R Support vector machines and evolutionary algorithms for classification: Single or together 2014 Springer Suykens JAK Signoretto M Argyriou A Regularization, optimization, kernels, and support vector machines 2015 CRC Press Vaishnavi VK Kuechler W Jr Design science research methods and patterns: Innovating information and communication technology 2008 Auerbach Publications Wang L Support vector machines: Theory and applications 2005 Springer Wiering MA Schomaker L Suykens JAK Signoretto M Argyriou A Multi-layer support vector machines Regularization, optimization, kernels, and support vector machines 2015 CRC Press 457 476 Yang, C., & Du, M. (2018). Booming online education brings chances, challenges in China. People's Daily. Retrieved from http://www.chinadaily.com.cn/edu/2018-02/08/content_35667621.htm

0

PMC9753873

NO-CC CODE

2022-12-16 23:26:20

no

Educ Inf Technol (Dordr). 2022 Dec 15;:1-15

utf-8

Educ Inf Technol (Dordr)

10.1007/s10639-022-11514-6

oa_other

==== Front Pediatr Res Pediatr Res Pediatric Research 0031-3998 1530-0447 Nature Publishing Group US New York 2419 10.1038/s41390-022-02419-8 Clinical Research Article Pre-pandemic support for shared reading buffers adverse parenting impacts: an RCT in Brazil Piccolo Luciane R. Luciane.Piccolo@nyulangone.org 1 Oliveira João B. A. 2 Hirata Guilherme 3 Canfield Caitlin F. 1 Roby Erin 1 Mendelsohn Alan L. 1 1 grid.137628.9 0000 0004 1936 8753 Department of Pediatrics, Division of Developmental and Behavioral Pediatrics, NYU Grossman School of Medicine, 462 First Ave—Bellevue Hospital, New York, NY 10016 USA 2 Instituto Alfa e Beto, 538 Lineu Anterino Mariano st, Uberlândia, MG 38402-346 Brazil 3 IDados, 470 Visconde de Pirajá st., Rio de Janeiro, RJ 22410-002 Brazil 15 12 2022 18 4 8 2022 5 11 2022 21 11 2022 © The Author(s), under exclusive licence to the International Pediatric Research Foundation, Inc 2022, Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. Background To examine whether (1) a parent-child reading program (Universidade do Bebê [UBB]), conducted in Brazil pre-pandemic can support parenting and parent-child reading 6 months into the pandemic, (2) cognitive stimulation at pandemic onset mediates effects of UBB on these outcomes, and (3) UBB pre-pandemic buffers associations between COVID-19-related distress and parenting/parent-child reading 6 months into the pandemic. Methods 400 women, either pregnant or with children 0–24 months, were randomized to UBB (n = 200) or control groups. UBB consisted of monthly parent workshops focusing on parent-child reading and a book-lending library. Assessments pre-pandemic (June-2019) and at pandemic onset (April-2020) included cognitive stimulation. Assessments 6 months into the pandemic (October-2020) included COVID-19 exposure/impact/distress, as well as parenting and parent-child reading. Results 133 families (n = 69 UBB) contributed data 6 months into the pandemic. Participation in UBB pre-pandemic was associated with parent-child reading but not parenting 6 months into the pandemic. Indirect effects of UBB through cognitive stimulation at pandemic onset were observed for both outcomes. Increased COVID-19-related distress was significantly associated with reduced parenting/parent-child reading 6 months into the pandemic in the control group only. Conclusion Promotion of cognitive stimulation pre-pandemic may have reduced risk for effects of the pandemic on parenting/parent-child reading. Clinical trial registration The trial has been registered with the Brazilian Clinical Trials Registry RBR-29RZDH on 05/28/2018. Impact This is the first study showing sustained impacts of a reading aloud intervention beginning in pregnancy and early infancy implemented pre-pandemic. Findings suggest that participation in a reading-aloud intervention buffered associations between COVID-19 distress and parenting/parent-child reading 6 months into the pandemic. Novel empirical evidence suggests that promotion of cognitive stimulation prior to the pandemic may buffer its impacts on parenting and parent-child book reading following onset in low- and middle-income countries. Findings provide important new support for implementation of parent-child reading aloud programs and likely have implications for early childhood development beyond the COVID-19 pandemic for disasters generally. ==== Body pmcIntroduction As with previous disasters,1–3 the COVID-19 pandemic may compound pre-existing stressors associated with poverty and exacerbate disparities in health and education, particularly in low- and middle-income countries (LMICs).4,5 It has been projected that approximately 10 million children are at high risk for early developmental delays (90% of whom are from LMICs) and long-term deficits in educational and professional achievement due to childcare disruption during the COVID-19 pandemic.4 Although these projections raise great concern, there has been limited research on strategies that may mitigate delays in early child development resulting from the pandemic, such as preventive parenting interventions, including those offered prior to the pandemic.6 Emerging studies have shown associations between COVID-19-related events (e.g., income loss, food insecurity, overcrowding, child care disruption),7,8 as well as parenting and parents’ mental health, particularly for families with infants and toddlers9–17 and those with limited pre-existing resources.13,15,18 For instance, parent stress related to the COVID-19 pandemic has been associated with harsh parenting19,20 and changes in parent-child reading and playing routines,21–23 resulting in adverse impacts on child cognitive-linguistic,14,24–28 and psychosocial29–31 development. These findings are consistent with conceptual models and pre-pandemic evidence suggesting that both exposure to stressors and limited resources are major contributors to disparities in child development.32–34 Specifically, this literature indicates that environmental stressors, similar to those related to the COVID-19 pandemic,9,15,20,35 are associated with psychosocial vulnerabilities and relational health (i.e., parent-child relationship quality and parenting practices) and that resource deprivation may limit provision of cognitively stimulating materials and experiences in the home.32–34 Prior to the COVID-19 pandemic, many preventive early childhood development (ECD) programs focused on promotion of parenting and cognitive stimulation as a key strategy to prevent effects of stressors on relational health and child development.34,36 For example, a reading aloud program called Universidade do Bebê (UBB), the focus of the current analysis, has undergone two studies in Brazil prior to the pandemic showing impacts on parenting (cognitive stimulation and quantity and quality of reading interactions) and child cognitive-linguistic and socioemotional outcomes in families with low income.37,38 Emerging studies have suggested that these strategies may also mitigate early learning losses resulting from the COVID-19 pandemic.39,40 Limited research on preventive programs focusing broadly on positive parenting and delivered pre-pandemic, such as Family Foundations (FF), has shown positive impacts on parenting and child behaviors in the early phase of the pandemic.6 However, there is no empirical evidence to support participation in pre-pandemic reading aloud programs as a buffer to the COVID-19 pandemic impacts on parenting practices and parent-child reading aloud. The lack of longitudinal data on young children41 (collected both pre- and post-pandemic) is also a barrier to making causal claims regarding impacts of interventions delivered prior to the pandemic and identifying potential protective factors.27 The current analysis seeks to address this limitation through longitudinal follow up of families in Brazil participating in a study of UBB, in which the program was delivered pre-pandemic.38 We have previously reported findings of this study through pandemic onset, including positive impacts on cognitive stimulation.38 Here, we seek to extend those findings by investigating: (1) whether UBB resulted in enhanced parenting and parent-child reading 6 months into the pandemic, (2) whether cognitive stimulation at pandemic onset and following termination of UBB mediated effects of UBB on parenting and parent-child book reading 6 months into the pandemic, and (3) whether UBB pre-pandemic buffered associations between COVID-19-related distress and parenting and parent-child reading 6 months into the pandemic. We hypothesized that participation in UBB pre-pandemic would have impacts on parenting and parent-child reading that would be sustained 6 months into the pandemic. We also hypothesized that this impact would be mediated by cognitive stimulation in the home at pandemic onset and that UBB would buffer negative effects of pandemic-related distress on parenting and parent-child reading aloud practices. Methods Design The current study was conducted in the context of a randomized controlled trial (RCT) investigating impacts of UBB on parenting and child outcomes in community centers in three neighborhoods in a city in northeast Brazil.38 This study was approved by the Ethics Committee of the Instituto de Medicina Integral Professor Fernando Figueira in Brazil under protocol number 2.503.697. The trial has been registered with the Brazilian Clinical Trials Registry RBR-29RZDH on 05/28/2018. All participants provided informed consent. Subjects Families were eligible for the study if the mother was pregnant (n = 66) or had children 0–24 months (0–12 months, n = 175; 12–24 months, n = 159), and met income criteria (less than half minimum wage per capita and total less than 3 times minimum wage per household) for a conditional federal cash-transfer program in Brazil (“Bolsa Família”).42 Families enrolled in the RCT were also eligible to participate in a lottery to a home visiting program, called “Programa Criança Feliz” (“Happy Child Program”; PCF).43 There were no exclusion criteria. Enrollment and the randomization processes were described in a previous publication.38 Four hundred families were randomized to UBB (n = 200) or control groups (n = 200) using a random number generator in Stata. Intervention The UBB intervention consisted of 1-h monthly parent workshops focused on parent-child shared reading that were led by a coach with a BA in psychology. The curricula included videos, live demonstrations, and practice of parent-child reading as well as discussion of strategies for reading with children at home and the importance of talking with children during reading, play, and daily routines. Parents were also encouraged to find a time to read with their child every day. Families in the UBB group also borrowed age-appropriate children’s books at each meeting. Each week between workshops, a staff member delivered and collected the books at families’ homes. The program was implemented from August 2019 to March 2020, at which time it was discontinued due to the pandemic.38 Procedures Figure 1 shows the study flow diagram and timeline of research activities. Pregnant women and families with children 0–24 months participated in in-person interviews in June 2019 and phone interviews in April 2020. At these time points, they provided data on sociodemographic characteristics and parent-child reading aloud and play routines at home by responding to surveys validated for the Brazilian population and used in previous studies.37,38 In October 2020, phone interviews were completed to assess COVID-19 exposure, impact, and related distress, as well as parenting practices and parent-child reading 6 months into the pandemic. The surveys and children’s direct assessments were conducted by research assistants (blind to study hypotheses and group assignment). Specialists in child development translated COVID-19 questionnaires and evaluated items’ relevance and semantic appropriateness for the Brazilian population.Fig. 1 Study flow diagram. Participant flow from eligibility assessment through enrollment, intervention, and data collection to completion of follow-up. Measures Outcome Parenting In a Pandemic Scale (PIPS; phone interviews in October 2020).44 This survey measures changes in parenting practices in terms of infection prevention, socioemotional support, and structured activities, including parent-child reading activities during the COVID-19 pandemic. It consists of 25 items using a 5-point Likert scale ranging from 0 (a lot less than before the pandemic) to 4 (a lot more than before the pandemic). Examples of items include “Read books with my child” and “Ensured that my child has good quality sleep (e.g., regular sleep and wake times, no screens in bed)”. In this study, PIPS total score and a composite of 4 questions about parent-child reading were analyzed as continuous variables (range 0–4; high scores indicate increased positive parenting practices or parent-child reading during the pandemic). In this sample, the overall scale (α = 0.78) and the parent-child reading composite (α = 0.88) presented good internal consistency. Predictors COVID-19 Exposure and Family Impact Survey (CEFIS; phone interviews in October 2020).45,46 The CEFIS measures levels of exposure to COVID-19 and its impact on families’ economic and psychosocial factors. Families were asked to respond to survey items considering events since March 2020. CEFIS has three domains: (1) Exposure, which consists of 25 Yes/No items corresponding to COVID-19-related events such as school closures, changes in employment, and exposure to the virus (scores 0–25); (2) Impact, which consists of 10 items rating COVID-19 impacts on family functioning factors using a 4-point Likert scale (1 = made it a lot better, 2 = made it a little better, 3 = made it a little worse, 4 = made it a lot worse, and a “not applicable” option); scale score is the average across items (range 1–4); and (3) Distress, which consists of 2 items measuring parents and children distress using a 10-point scale (0 = no distress to 10 = extreme distress); scale score is the average across items (range 0–10). High scores indicate high COVID-19 exposure/impact/distress. In this sample, the three subscales showed good (α = 0.75 for Exposure, α = 0.81 for Impact, and α = 0.76 for Distress) internal consistency. Mediator StimQ (phone interviews in April 2020).47 The StimQ measures cognitive stimulation in the home through parent-child interactions in play, shared reading, teaching, and daily routines. We used the StimQ Core subscales, which include: (1) frequency and quality of reading interactions (READ; scores 0–13); (2) caregiver-child verbal interactions (Parental Verbal Responsivity [PVR]; scores 0–14); and (3) caregiver teaching and play activities (Parent Involvement in Developmental Advance [PIDA]; scores 0–10). In this study, the total score was analyzed. The Brazilian version showed high internal consistency (α = 0.95) in a previous study.38 Covariates (in-person parent-surveys in June 2019) Child characteristics included child’s age, sex, and birth order. Family characteristics included the mother’s age, education (dichotomized as high school graduate or less), marital status, as well as household food insecurity and overcrowding (persons per room). Maternal depression was measured by the Edinburgh Postnatal Depression Scale (EPDS)48 and dichotomized using total score ≥ 10 as a cutoff.48,49 In addition, participation in the PCF home visiting program was scored dichotomously as 0 (control) or 1 (PCF offered by the municipality). Data analysis Analyses were based on intent-to-treat. Descriptive statistics were used to summarize participants’ sociodemographic characteristics and to describe COVID-19 pandemic exposure and impacts on low-income families’ well-being in northeast Brazil. Comparisons between the full and analytic samples as well as between randomization groups were conducted using chi-square (for dichotomous variables) and t-tests (for continuous indicators). To address our first aim, we performed t-test and linear regression analyses (adjusted for covariates and CEFIS scores) to investigate whether UBB conducted pre-pandemic may support parenting and parent-child reading 6 months into the pandemic (PIPS). For our second aim, we conducted mediation analysis using structural equation modeling (SEM) to understand whether cognitive stimulation (StimQ) following termination of UBB at pandemic onset mediates effects of UBB on parenting and parent-child reading 6 months into the pandemic (PIPS). Separate models for each of the dependent variables (i.e., parenting and parent-child reading - PIPS) were tested, adjusting for baseline covariates and cognitive stimulation (StimQ), as well as COVID-19 distress/impact/exposure scores (CEFIS). The “estat teffects” command in Stata was used to determine significance of all indirect effects for each model. For our third aim, moderation analyses were performed to investigate whether UBB pre-pandemic may buffer associations between COVID-19 distress/impact/exposure (independent variable; CEFIS) and parenting and parent-child reading during the pandemic (dependent variables; PIPS) while controlling for covariates. Subgroup analyses were conducted to examine associations between CEFIS scores and parenting and parent-child reading (PIPS) by randomization group, when the interaction term UBB*CEFIS was significant. In addition, we replicated the analyses by using multiple imputed data sets. Missing values of StimQ at pandemic onset (n = 114; UBB n = 50), CEFIS (n = 267; UBB n = 131), and PIPS (n = 267; UBB n = 131) were replaced using multiple imputation by randomization groups50 and a confirmatory intent-to-treat analysis was conducted. Table 1 shows results before imputation (results after imputation are shown in the Supplementary Materials).Table 1 Sample characteristics. Full sample (N = 400) Analytic sample (n = 133) pa Randomization Group pb Control (n = 64) UBB (n = 69) Child characteristics at baseline  Child’s Age in Months, mean (SD) 11.6 (6.7) 10.7 (6.6) 0.09 10.3 (6.5) 11.0 (6.7) 0.56  Child Sex – Female, % 48.8 50.8 0.91 45.3 53.6 0.34  First Born Child, % 35.5 36.7 0.65 31.2 39.1 0.34 Family characteristics at baseline  Mother’s Age in Years, mean (SD) 27.3 (6.6) 27.7 (5.8) 0.33 28.5 (6.4) 27.0 (5.2) 0.13  Mother High School Graduate, % 45.5 51.3 0.40 52.0 60.0 0.35  Parents Married or Living with a Partner, % 67.9 70.8 0.14 73.4 72.4 0.90  Depression, %c 35.3 33.8 0.66 29.7 37.7 0.33  Food Insecure, % 73.0 82.0 0.30 81.2 76.8 0.50  Overcrowding, mean (SD) 0.9 (0.3) 0.8 (0.3) 0.13 0.8 (0.3) 0.8 (0.3) 0.61  Offered PCF, % 50.2 48.1 0.62 51.6 44.9 0.45 CEFIS 6 months into the pandemic  Exposure, mean (SD) – 10.1 (3.8) – 10.0 (4.1) 10.2 (3.4) 0.79  Impact, mean (SD) – 2.4 (0.5) – 2.4 (0.5) 2.5 (0.4) 0.17  Distress, mean (SD) – 6.1 (2.6) – 5.9 (2.7) 6.2 (2.5) 0.48 CEFIS COVID-19 exposure and impact survey. ap value for comparisons between Full and Analytic Sample (t-test for continuous variables and chi-square for categorical variables). bp value for comparisons between Randomization Groups in the Analytic Sample (t-test for continuous variables and chi-square for categorical variables). cMet criteria if EPDS ≥ 10. Results Sample characteristics Enrollment took place from October 2018 through May 2019. The analytic sample consisted of 133 parents (n = 69 UBB) who responded to phone interviews in April and October 2020 with children 9.3 to 40.5 months (M = 24.8 months, SD = 8.1) at the time of the COVID-19 specific assessment in October 2020 (Fig. 1). The full (N = 400) and analytic (n = 133) samples had comparable sociodemographic characteristics at baseline (Table 1). In addition, sociodemographic characteristics of families lost to follow-up were comparable between randomization groups (Supplementary Materials, Table S1). Randomization groups did not differ in terms of exposure and impact of COVID-19. Table 1 shows that families in this study were exposed to an average of 10.1 events related to the pandemic (CEFIS Exposure subscale). In terms of impacts, the mean score on CEFIS Impact subscale was close to the mid-point (2.5), indicating that overall families’ well-being had not been significantly affected by the pandemic. For families in both groups, distress levels (CEFIS Distress subscale) were above the mean point (5). Similar results were found after imputation (see Supplementary Materials, Table S2). Aim 1. Association between participation in UBB pre-pandemic and parenting and parent-child reading 6 months into the pandemic Differences between UBB and control groups were significant for parent-child reading (t(131) = 2.47, p = 0.01; Cohen’s d = 0.36) but not for parenting (t(130) = 1.07, p = 0.28; d = 0.19). Models adjusted for covariates showed that UBB was associated with parent-child reading (β = 0.22, p = 0.03) but not overall parenting (β = 0.06, p = 0.33) 6 months into the pandemic (Fig. 2). The association between UBB and parent-child reading 6 months into the pandemic was retained after adjusting for CEFIS Distress (β = 0.22, p = 0.04), Impact (β = 0.25, p = 0.01) or Exposure (β = 0.22, p = 0.03). Results were similar after data imputation (see Supplementary Materials).Fig. 2 Effects of UBB. Effects of UBB on a parenting and b parent–child reading 6 months into the pandemic, including total, direct, and indirect effects mediated by impacts on cognitive stimulation in the home at pandemic onset. β = standardized coefficients. Models were adjusted for baseline covariates and cognitive stimulation, as well as COVID-19-related Distress. Aim 2. Cognitive stimulation at pandemic onset mediates effects of UBB on parenting and parent-child book reading 6 months into the pandemic Indirect effects of UBB through cognitive stimulation at pandemic onset were observed for both parenting and parent-child book reading 6 months into the pandemic, in models adjusted for baseline covariates and cognitive stimulation as well as CEFIS Distress (Fig. 2). Results were similar when the models were adjusted for CEFIS Impact or CEFIS Exposure and when missing values were imputed (see Supplementary Materials, Table S3). Aim 3. UBB pre-pandemic buffers associations between COVID-19 related distress and parenting and parent-child reading 6 months into the pandemic Main effects of CEFIS Distress were significant for parenting (β = −0.27, p = 0.01) and parent-child reading (β = −0.25, p = 0.03). In addition, main effects of CEFIS Impact were observed for parenting (β = −0.28, p = 0.01) and parent-child reading (β = −0.34, p = 0.002) 6 months into the pandemic. There were no significant main effects of CEFIS Exposure. Results were similar after data imputation (see Tables S3 and S4 in Supplementary Materials). A significant interaction was found for UBB and CEFIS Distress for both overall parenting (β = 0.32, p = 0.024) and parent-child reading (β = 0.35, p = 0.020) 6 months into the pandemic. Figure 3 shows that negative associations between COVID-19 Distress scale and parenting/parent-child reading were buffered for the UBB group (parenting: β = −0.02, p = 0.91; parent-child reading: β = −0.14, p = 0.29), but were significant for the control group (parenting: β = −0.34, p = 0.014; parent-child reading: β = −0.45, p = 0.002). That is, increased COVID-19 Distress was significantly associated with reduced positive parenting and parent-child reading 6 months into the pandemic in the control group only. Similarly, a significant interaction was also found for UBB and CEFIS Impact for overall parenting (β = 0.39, p = 0.003) and parent-child book reading (β = 0.36, p = 0.006) 6 months into the pandemic. Negative associations between COVID-19 Impact scale and parenting/parent-child reading were buffered for the UBB group (parenting: β = −0.13, p = 0.44; parent-child reading: β = −0.16, p = 0.20), but were significant for the control group (parenting: β = −0.40, p = 0.01; parent-child reading: β = −0.35, p = 0.03). There were no significant interactions for UBB and CEFIS Exposure (parenting: β = −0.11, p = 0.38; parent-child reading: β = −0.07, p = 0.61).Fig. 3 UBB Moderated effects of COVID-19-related distress level on parenting and parent–child book reading 6 months into the pandemic. a, b The negative associations between COVID-19-related stress and parenting 6 months into the pandemic were buffered for the UBB group, but statistically significant for the control group. Models were adjusted for baseline covariates. For brevity, we only illustrate the significant interactions between CEFIS Distress and UBB (Fig. 3). Results for all outcomes, including after imputation, are presented in Supplementary Materials (Tables S4 and S5). Discussion This study demonstrated that (1) UBB showed sustained increased parent-child reading 6 months into the pandemic, (2) effects of UBB on parenting and parent-child book reading 6 months into the pandemic were mediated by cognitive stimulation at pandemic onset, and (3) participation in UBB pre-pandemic buffered links between COVID-19 related distress/impact and parenting practices and parent-child reading 6 months into the pandemic. Findings reinforce the importance of research on and implementation of parenting programs to support vulnerable populations, as they may not only address disparities, but prevent the exacerbation of such inequality during disasters, globally and particularly in LMICs.2,3 This study revealed a number of important results. First, our analyses demonstrate that participation in UBB beginning in pregnancy and early infancy, prior to the COVID-19 pandemic was associated with parent-child reading 6 months into the pandemic. This novel finding supports implementation of preventive programs focusing on parent-child reading aloud and contributes to limited literature6 demonstrating how engaging in parenting interventions pre-pandemic showed benefits later during the COVID-19 pandemic. Second, findings extend prior work showing the effects of UBB on cognitive stimulation38 and demonstrate that these early impacts also mediate effects of UBB on parenting and parent-child reading, with effects sustained 6 months into the pandemic. These findings align with existing conceptual models,32,34 and suggest that promotion of parent-child reading and provision of books beginning in pregnancy and early infancy supports early relational health and childhood development. Finally, this study showed that significant effects of adverse childhood experiences (ACEs) associated with COVID-197,8,51 on parenting and parent-child reading were buffered by participation in a parent-child reading aloud program pre-pandemic. Research suggests that children who experience ACEs and are at greater risk for developmental problems may nonetheless thrive when they also experience positive parent-child interactions and relationships.52 These findings are clinically important given that the pandemic and other natural disasters may have unique impacts on children’s development as they not only exacerbate existing disparities and make ACEs more likely, but may also create barriers to positive childhood experiences and positive relational health that are known to support flourishing.1–3,52 The current findings extend these results to ACEs associated with the COVID-19 pandemic,51 and provide further evidence in support of recent American Academy of Pediatrics (AAP) statements on the importance of promotion of positive relational health.36 Importantly, these results also highlight modifiable protective mechanisms for children’s potential learning losses that are posited to result from the COVID-19 pandemic.4,40 In addition, findings add to existing evidence of supporting a role for programs targeting positive parenting activities such as reading aloud and play (e.g., “Reach Out and Read” and “Video Interaction Project”) in the context of traumatic events broadly. For example, a study in the Philippines suggested that support for reading aloud helped buffer experience of trauma following Typhoon Haiyan.53 This study used data from an RCT of parents of infants and toddlers in a LMIC to demonstrate strong empirical evidence of the impacts of a reading aloud program delivered pre-pandemic on parenting and parent-child reading 6 months into the pandemic. However, there are a number of limitations. First, validated measures to evaluate COVID-19 exposure/impacts and parenting practices in the context of the pandemic in Brazil were unavailable. We addressed this limitation by examining psychometric characteristics of the translated scales, which demonstrated good internal consistency. Second, approximately a third of the original cohort was interviewed for the assessment that took place 6 months following pandemic onset. However, comparison of sociodemographic characteristics did not show differences between the analytic and full samples. Further, analyses utilizing data imputation methods had comparable findings, with significant differences retained for all measures. Third, given that perceived pandemic-related distress may vary depending on the timing of the evaluation,54 interpretations of this study’s results may be restricted to the time-point when the data was collected (October 2020), which is considered the end of the first wave of the COVID-19 pandemic in Brazil.55 Fourth, further study is needed to determine whether findings generalize to other regions in Brazil, LMICs or high-income countries. Fifth, additional information about siblings (e.g., ages, school grades) was not available. This is an important limitation as older siblings in the household may have influenced parenting and reading routines in the context of homeschooling during the pandemic.56 To further examine the potential role of siblings, we re-ran all analyses using number of siblings, rather than first-born, as a covariate. Findings were similar in terms of statistical significance and effect sizes. Sixth, we did not use diaries to examine parent-child reading. Instead, we used parent surveys, similar to other studies during the pandemic.22,23 Although our prior work has documented comparable findings when using StimQ and diaries,57 no studies to our knowledge have examined parent-child reading through parents’ diaries during the pandemic. Finally, this study’s results allow only indirect evidence for the potential buffering effect of UBB on children’s COVID-related learning losses. Future studies should investigate whether programs that vary in intensity and focus might have similar buffering effects in the context of crises. In addition, such studies should investigate whether intervention impacts on early childhood development and families’ wellbeing might differ depending on location, level of exposure to traumatic events, race/ethnicity, and existing family strengths/challenges. Conclusion This study indicated that promotion of cognitive stimulation in the home through parent-child reading pre-pandemic may be a useful strategy for buffering negative effects of COVID-19 on parenting and parent-child reading, with potential for preventing delays in early child development. These findings have implications for the design and implementation of preventive programs to support vulnerable families, within and beyond the COVID-19 pandemic context. Supplementary Information Checklist item Supplementary Material Supplementary information The online version contains supplementary material available at 10.1038/s41390-022-02419-8. Acknowledgements We would like to thank Dr. Adriana Weisleder (Northwestern University) for contributing to the design of the study in Brazil. We also would like to thank Mariana Leite (IDados) for providing technical support to this study, Walfrido Duarte Neto and Danubia Santos (Instituto Alfa e Beto) for supervising fieldwork and coordinating parent workshops, as well as the intervention and evaluation teams for making the study possible. We are also thankful to the local authorities, as well as to the participating families. Author contributions L.R.P. drafted the initial manuscript, conceptualized and designed the study, designed the data collection instruments, conducted statistical analyses, and reviewed and revised the manuscript; J.B.A.O. conceptualized and designed the study, coordinated the implementation of the intervention in Brazil, and critically reviewed the manuscript; G.H. coordinated and supervised recruitment, enrollment, randomization, as well as data collection, conducted statistical analyses, and reviewed and revised the manuscript; C.F.C. and E.R. conceptualized and designed the study, and reviewed and revised the manuscript; A.L.M. conceptualized and designed the study, developed the intervention, designed the data collection instruments, and reviewed and revised the manuscript. All authors approved the final manuscript as submitted. Funding Instituto Alfa e Beto. Data availability The datasets generated and/or analyzed during the current study are available from the corresponding author on reasonable request. Competing interests The authors declare no competing interests. Informed consent Informed consent was obtained from the pregnant women and parents of the children at enrollment. Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References 1. Kammerbauer M Wamsler C Social inequality and marginalization in post-disaster recovery: challenging the consensus? Int. J. Disaster Risk Reduct. 2017 24 411 418 10.1016/j.ijdrr.2017.06.019 2. do Carmo RF Silva Júnior JVJ Pastor AF de Souza CDF Spatiotemporal dynamics, risk areas and social determinants of dengue in Northeastern Brazil, 2014–2017: an ecological study Infect. Dis. Poverty 2020 9 153 10.1186/s40249-020-00772-6 33143752 3. Diderichsen F Augusto LG da S Perez B Understanding social inequalities in Zika infection and its consequences: a model of pathways and policy entry-points Glob. Public Health 2019 14 675 683 10.1080/17441692.2018.1532528 30301438 4. McCoy DC Global estimates of the implications of COVID‐19‐related preprimary school closures for children’s instructional access, development, learning, and economic wellbeing Child Dev. 2021 92 13658 10.1111/cdev.13658 5. Osendarp S The COVID-19 crisis will exacerbate maternal and child undernutrition and child mortality in low- and middle-income countries Nat. Food 2021 27 2021 2 476 484 6. Feinberg ME Building long-term family resilience through universal prevention: 10-year parent and child outcomes during the COVID-19 pandemic Fam. Process 2022 61 76 90 10.1111/famp.12730 34927239 7. Patrick SW Well-being of parents and children during the COVID-19 pandemic: a national survey Pediatrics 2020 146 e2020016824 10.1542/peds.2020-016824 32709738 8. Gagné M-H Piché G Clément M-È Villatte A Families in confinement: a pre–post COVID-19 study Couple Fam. Psychol. Res. Pract. 2021 10 260 270 10.1037/cfp0000179 9. Lucassen N de Haan AD Helmerhorst KOW Keizer R Interrelated changes in parental stress, parenting, and coparenting across the onset of the COVID-19 pandemic J. Fam. Psychol. 2021 35 1065 1076 10.1037/fam0000908 34398624 10. Kahn M Barnett N Glazer A Gradisar M COVID-19 babies: auto-videosomnography and parent reports of infant sleep, screen time, and parent well-being in 2019 vs 2020 Sleep. Med. 2021 85 259 267 10.1016/j.sleep.2021.07.033 34388504 11. Layton, H., Owais, S., Savoy, C. D. & Van Lieshout, R. J. Depression, anxiety, and mother-infant bonding in women seeking treatment for postpartum depression before and during the COVID-19 pandemic. J. Clin. Psychiatry 82, 21m13874 (2021). 10.4088/JCP.21m13874 12. Mayopoulos GA COVID-19 is associated with traumatic childbirth and subsequent mother-infant bonding problems J. Affect. Disord. 2021 282 122 125 10.1016/j.jad.2020.12.101 33412491 13. Hamadani JD Immediate impact of stay-at-home orders to control COVID-19 transmission on socioeconomic conditions, food insecurity, mental health, and intimate partner violence in Bangladeshi women and their families: an interrupted time series Lancet Glob. Heal. 2020 8 e1380 e1389 10.1016/S2214-109X(20)30366-1 14. Provenzi, L. et al. Prenatal maternal stress during the COVID-19 pandemic and infant regulatory capacity at 3 months: a longitudinal study. Dev. Psychopathol. 1–9 (2021). 10.1017/S0954579421000766 15. Taubman – Ben-Ari O Ben-Yaakov O Chasson M Parenting stress among new parents before and during the COVID-19 pandemic Child Abus. Negl. 2021 117 105080 10.1016/j.chiabu.2021.105080 16. Fernandes DV Canavarro MC Moreira H The role of mothers’ self‐compassion on mother–infant bonding during the COVID‐19 pandemic: a longitudinal study exploring the mediating role of mindful parenting and parenting stress in the postpartum period Infant Ment. Health J. 2021 42 621 635 10.1002/imhj.21942 34407224 17. Kimura M Kimura K Ojima T Relationships between changes due to COVID-19 pandemic and the depressive and anxiety symptoms among mothers of infants and/or preschoolers: a prospective follow-up study from pre-COVID-19 Japan BMJ Open 2021 11 e044826 10.1136/bmjopen-2020-044826 18. Prime H Wade M Browne DT Risk and resilience in family well-being during the COVID-19 pandemic Am. Psychol. 2020 75 631 643 10.1037/amp0000660 32437181 19. Fosco GM Sloan CJ Fang S Feinberg ME Family vulnerability and disruption during the COVID‐19 pandemic: prospective pathways to child maladjustment J. Child Psychol. Psychiatry 2022 63 47 57 10.1111/jcpp.13458 34101185 20. Sari NP van IJzendoorn MH Jansen P Bakermans-Kranenburg M Riem MME Higher levels of harsh parenting during the COVID-19 lockdown in the Netherlands Child Maltreat 2022 27 156 162 10.1177/10775595211024748 34134541 21. López-Escribano C Escudero A Pérez-López R An exploratory study about patterns of parental home literacy activities during the COVID-19 confinement among spanish families Early Educ. Dev. 2021 32 812 829 10.1080/10409289.2021.1916184 22. Wheeler, D. L. & Hill, J. C. The impact of COVID-19 on early childhood reading practices. J. Early Child. Lit. 1–20 (2021). 10.1177/14687984211044187 23. Read, K., Gaffney, G., Chen, A. & Imran, A. The impact of COVID-19 on families’ home literacy practices with young children. Early Child. Educ. J. 1–10 (2021). 10.1007/s10643-021-01270-6 24. Adegboye D Understanding why the COVID‐19 pandemic‐related lockdown increases mental health difficulties in vulnerable young children JCPP Adv. 2021 1 e12005 10.1111/jcv2.12005 34485985 25. Russell BS Hutchison M Tambling R Tomkunas AJ Horton AL Initial challenges of caregiving during COVID-19: caregiver burden, mental health, and the parent–child relationship Child Psychiatry Hum. Dev. 2020 51 671 682 10.1007/s10578-020-01037-x 32749568 26. Moore SA Impact of the COVID-19 virus outbreak on movement and play behaviours of Canadian children and youth: a national survey Int. J. Behav. Nutr. Phys. Act. 2020 17 85 10.1186/s12966-020-00987-8 32631350 27. Spinelli M Lionetti F Pastore M Fasolo M Parents’ stress and children’s psychological problems in families facing the COVID-19 outbreak in Italy Front. Psychol. 2020 11 1713 10.3389/fpsyg.2020.01713 32719646 28. González M School readiness losses during the COVID-19 outbreak: a comparison of two cohorts of young children Child Dev. 2022 93 910 924 10.1111/cdev.13738 35194777 29. Spinelli M Parenting stress during the COVID-19 outbreak: Socioeconomic and environmental risk factors and implications for children emotion regulation Fam. Process 2021 60 639 653 10.1111/famp.12601 32985703 30. Babore, A. et al. Mothers’ and children’s mental health during the COVID-19 pandemic lockdown: the mediating role of parenting stress. Child Psychiatry Hum. Dev. 1–13 (2021). 10.1007/s10578-021-01230-6 31. Achterberg M Dobbelaar S Boer OD Crone EA Perceived stress as mediator for longitudinal effects of the COVID-19 lockdown on wellbeing of parents and children Sci. Rep. 2021 11 2971 10.1038/s41598-021-81720-8 33536464 32. Wolf S McCoy DC Household socioeconomic status and parental investments: direct and indirect relations with school readiness in Ghana Child Dev. 2019 90 260 278 10.1111/cdev.12899 28708296 33. Masarik AS Conger RD Stress and child development: a review of the Family Stress Model Curr. Opin. Psychol. 2017 13 85 90 10.1016/j.copsyc.2016.05.008 28813301 34. Shaw DS Mendelsohn AL Morris PA Reducing poverty-related disparities in child development and school readiness: the Smart Beginnings tiered prevention strategy that combines pediatric primary care with home visiting Clin. Child Fam. Psychol. Rev. 2021 24 669 683 10.1007/s10567-021-00366-0 34505232 35. Westrupp, E. M. et al. Child, parent, and family mental health and functioning in Australia during COVID-19: comparison to pre-pandemic data. Eur. Child Adolesc. Psychiatry 1–14 (2021). 10.1007/s00787-021-01861-z 36. Garner A Yogman M Preventing childhood toxic stress: Partnering with families and communities to promote relational health Pediatrics 2021 148 e2021052582 10.1542/peds.2021-052582 34312296 37. Weisleder, A. et al. Reading aloud and child development: A cluster-randomized trial in Brazil. Pediatrics 141, e20170723 (2018). 10.1542/peds.2017-0723 38. Piccolo, L. R., Oliveira, J. B. A., Hirata, G., Duarte Neto, W. & Mendelsohn, A. L. Supporting reading aloud beginning prenatally and in early infancy: a randomized trial in Brazil. J. Dev. Behav. Pediatr. 43, e590–e597 (2022). 10.1097/DBP.0000000000001118 39. Nuryanti N Iswara PD Home literacy environment: the solution to improve early reading skills of students in primary school during COVID-19 Int. Conf. … 2021 3 219 228 40. Bao X Qu H Zhang R Hogan TP Modeling reading ability gain in kindergarten children during COVID-19 school closures Int. J. Environ. Res. Public Health 2020 17 1 13 10.3390/ijerph17176371 41. Racine N Global prevalence of depressive and anxiety symptoms in children and adolescents during COVID-19 JAMA Pediatr. 2021 175 1142 10.1001/jamapediatrics.2021.2482 34369987 42. Rasella D Aquino R Santos CAT Paes-Sousa R Barreto ML Effect of a conditional cash transfer programme on childhood mortality: a nationwide analysis of Brazilian municipalities Lancet 2013 382 57 64 10.1016/S0140-6736(13)60715-1 23683599 43. Girade, H. A. ‘Criança Feliz’: A programme to break the cycle of poverty and reduce the inequality in Brazil. (2018). 44. Waller, R., Chester, M., Rodriguez, Y. & Wagner, N. Development of the Parenting In a Pandemic Scale (PIPS). (2020). 10.31234/osf.io/f8tzm 45. Enlow PT Validation of the COVID-19 exposure and family impact scales J. Pediatr. Psychol. 2022 47 259 269 10.1093/jpepsy/jsab136 34969064 46. Kazak AE COVID-19 exposure and family impact scales: factor structure and initial psychometrics J. Pediatr. Psychol. 2021 46 504 513 10.1093/jpepsy/jsab026 33749794 47. Dreyer BP Mendelsohn AL Tamis-LeMonda CS Assessing the child’s cognitive home environment through parental report; reliability and validity Early Dev. Parent. 1996 5 271 287 10.1002/(SICI)1099-0917(199612)5:4<271::AID-EDP138>3.0.CO;2-D 48. Santos IS Comparing validity of Edinburgh scale and SRQ20 in screening for post-partum depression Clin. Pract. Epidemiol. Ment. Heal. 2007 3 18 10.1186/1745-0179-3-18 49. Cox JL Holden JM Sagovsky R Detection of postnatal depression: development of the 10-item Edinburgh postnatal depression scale Br. J. Psychiatry 1987 150 782 786 10.1192/bjp.150.6.782 3651732 50. Sullivan TR White IR Salter AB Ryan P Lee KJ Should multiple imputation be the method of choice for handling missing data in randomized trials? Stat. Methods Med. Res. 2018 27 2610 10.1177/0962280216683570 28034175 51. Jiao WY Behavioral and emotional disorders in children during the COVID-19 epidemic J. Pediatr. 2020 221 264 266.e1 10.1016/j.jpeds.2020.03.013 32248989 52. Bethell CD Gombojav N Whitaker RC Family resilience and connection promote flourishing among US children, even amid adversity Health Aff. 2019 38 729 737 10.1377/hlthaff.2018.05425 53. Agustin, M. S., Ramos-Bonoan, C., Lorenzana, R., Klass, P. & Needlman, R. Picture books and reading aloud to support children after a natural disaster: an exploratory study Int. J. Emerg. Ment. Heal. Hum. Resil. 21, 1–6 (2018). 54. Park CL Psychological resilience early in the COVID-19 pandemic: stressors, resources, and coping strategies in a national sample of Americans Am. Psychol. 2021 76 715 728 10.1037/amp0000813 34081505 55. Zeiser FA First and second COVID-19 waves in Brazil: A cross-sectional study of patients’ characteristics related to hospitalization and in-hospital mortality Lancet Reg. Heal. - Am. 2022 6 100107 56. Sun X Implications of COVID-19 school closures for sibling dynamics among U.S. Latinx children: a prospective, daily diary study Dev. Psychol. 2021 57 1708 1718 10.1037/dev0001196 34807691 57. Mendelsohn AL Primary care strategies for promoting parent-child interactions and school readiness in at-risk families Arch. Pediatr. Adolesc. Med. 2011 165 33 41 10.1001/archpediatrics.2010.254 21199978

36522551

PMC9753875

NO-CC CODE

2022-12-16 23:26:20

no

Pediatr Res. 2022 Dec 15;:1-8

utf-8

Pediatr Res

10.1038/s41390-022-02419-8

oa_other

==== Front Food Ethics Food Ethics Food Ethics 2364-6853 2364-6861 Springer International Publishing Cham 114 10.1007/s41055-022-00114-2 Research Article Values of Australian Meat Consumers Related to Sheep and Beef Cattle Welfare: What Makes a Good Life and a Good Death? http://orcid.org/0000-0001-7073-5588 Buddle Emily A. emily.buddle@adelaide.edu.au 1 http://orcid.org/0000-0002-9435-8876 Bray Heather J. 12 http://orcid.org/0000-0002-1547-6031 Ankeny Rachel A. 1 1 grid.1010.0 0000 0004 1936 7304 Food Values Research Group, School of Humanities, The University of Adelaide, 5005 Adelaide, South Australia Australia 2 grid.1012.2 0000 0004 1936 7910 School of Biological Sciences, The University of Western Australia, 6009 Perth, Western Australia Australia 15 12 2022 2023 8 1 55 11 2022 © The Author(s), under exclusive licence to Springer Nature Switzerland AG 2022, Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. There has been growing global interest in livestock animal welfare. Previous research into attitudes towards animal welfare has focused on Europe and the United States, with comparatively little focus on Australia, which is an important location due to the prominent position of agriculture economically and culturally. In this article, we present results from qualitative research on how Australian meat consumers conceptualise sheep and beef cattle welfare. The study was conducted in two capital cities (Melbourne, Victoria and Adelaide, South Australia) and a much smaller rural centre (Toowoomba, Queensland) using focus groups (involving 40.9% of participants) and mall-intercept interviews (59.1% of participants), totalling 66 participants. Qualitative analysis highlights that participants had clear ideas of what it means for an animal to live a ‘good life’ and experience a ‘good death,’ with their beliefs strongly tied to their expectations and cultural understandings of what Australian agriculture ‘should be.’ In response to open-ended questions, participants expressed attitudes that relied on romanticised visions of the ‘rural idyll’ as seen in frequent discussions about what is ‘normal’ for sheep meat and beef production, and relatedly, what count as ‘natural behaviours.’ Many participants rejected anything associated with the ‘other,’ classifying it as not ‘normal’: we argue that which is not considered normal, including intensive production, foreign ownership, and halal slaughter practices, appear to place participants’ conceptualizations of an animal’s ‘good death,’ and in turn the potential for a ‘good life,’ at risk. Keywords Livestock animal welfare Australia Meat Slaughter Natural Australian Research Council LP130100419 Ankeny Rachel A. issue-copyright-statement© Springer Nature Switzerland AG 2023 ==== Body pmcIntroduction Debates about what makes food ‘good’ have been occurring in both scholarly and public domains in recent years (Ankeny 2012; Lewis and Huber 2015; Wilkerson 2016). Increasingly, good food is viewed not just as being safe, tasty, and nutritious, but also as being produced in a way that is considered good by society. Growing public interest in the welfare of food production animals has been attributed to increasing awareness of animal sentience (Broom 2014) and to increasing intensification of animal agriculture over recent decades (Buddle et al. 2018a). A range of animal welfare-related claims on meat products (Bray and Ankeny 2017; Malek et al. 2019) have now enabled consumers to ‘vote with their forks’ by boycotting certain products and ‘buycotting’ others (Michelletti 2011). In addition, food production methods are increasingly scrutinised by the media (Phillipov 2016; Carey et al. 2017; Sinclair et al. 2018) and by animal welfare activists via social media (Rodan and Mummery 2014; Buddle et al. 2017, 2018a). Although community understandings of farm animal welfare have been extensively examined in Europe (Van Pouke et al. 2006; María 2006; Vanhonacker and Verbeke 2009; Vanhonacker et al. 2010; Verbeke et al. 2010), North America (Spooner et al. 2014; Muringai et al. 2017), and Latin America (Miranda-de La Lama et al. 2017; Vargas-Bello-Pérez et al. 2017), there has been comparatively less research in Australia (see Future Eye 2018 for an Australian Government-commissioned report). Recent media attention on animal production ethics (Buddle and Bray 2019), particularly the live export of sheep and beef cattle (Sinclair et al. 2018), and the labelling of animal products as free-range (Carey et al. 2017), indicates that public attention to animal welfare issues has been steadily growing in Australia. Although many Australians are removed from the production of food (over 80% of people live in major cities, see Australian Government 2015), the Australian red meat sector is a “significant contributor to the rural economy,” with the total value of Australia’s beef and sheep meat industries estimated to be AUD$17 billion annually pre-COVID-19 (Meat and Livestock Australia 2018, para. 4). Australia has one of the highest per capita meat consumption rates in the world (approximately 111 kg of meat per person annually: Australian Bureau of Agricultural and Resource Economics and Sciences 2016) with meat eating having long been considered essential to both Australian meals and identity (Santich 1995, 2014; Ankeny 2008; Chen 2016a). Furthermore, biosecurity restrictions on imported animal products have led to reliance on domestic production of cattle and sheep-meat mostly by grazing on native and improved pastures. Australia’s climate means that few sheep and beef cattle are housed; although the use of feedlots is common, producing 80% of beef that is sold through domestic supermarkets, cattle on average only spend around 10 to 15 per cent of their lives in a feedlot and the rest on grass (Salvin et al. 2020). A near duopoly in the retail sector gives retailers an arguably greater role than producers or consumers in how food becomes valued by consumers (Dixon 2003; Phillipov 2016, 2017). The significance of red meat production and consumption in Australia makes it distinct in critical ways as compared to systems in the Northern Hemisphere, and hence it is important to understand community attitudes to beef and sheep meat production in Australia. Research on Australian community understandings of farm animal welfare has generally relied on surveys and knowledge-based assessments regarding specific practices (e.g., Coleman et al. 2016). In contrast, we examine Australian meat consumers’ understandings of what constitutes a ‘good’ life and a ‘good’ death for beef cattle and meat sheep, and explore how perceptions of animal welfare are culturally constructed in an Australian context. Our research approach is grounded in social constructivism, which explores subjective meanings that are “formed through interaction with others…and through historical and cultural norms that operate in individuals’ lives” (Creswell 2013, p. 25). Qualitative research methods can also highlight underlying motivations, values, or attitudes about an issue which cannot be revealed through closed-ended surveys (Malhotra 2006). Materials and methods This research was approved by the University of Adelaide’s Human Research Ethics Committee (H-2018-210) and conducted in accordance with the Australian national guidelines (National Health and Medical Research Council 2007, updated 2018). The research was conducted in Melbourne, Victoria (population of approximately 4.65 million); Adelaide, South Australia (population of approximately 1.2 million); and Toowoomba, Queensland (population of approximately 115,000). These locales were selected to capture understandings of animal welfare from a large capital city (Melbourne), a smaller capital city (Adelaide), and a regional centre (Toowoomba) to explore potential differences between populations and experiences. Consistent with qualitative approaches (Denzin and Lincoln 2011), this research used focus groups and ‘mall-intercept’ interviews (Bush and Hair 1985), with the latter employed to provide more balance to the overall sample in terms of demographics, particularly socioeconomic status. Three focus groups were conducted with 9 participants per group (27 participants or 40.9% in total) while 39 people (59.1% of participants) were involved in mall-intercept interviews. Participants for focus groups were recruited through community announcements and flyers distributed at public events such as university open days, and through social media including Facebook and Twitter. Only those over the age of 18 who identified as meat consumers and who spoke English were eligible to participate, with good balance of various demographics achieved by combining the focus group and mall-intercept interview methodologies. Focus groups and interviews were conducted using a semi-structured script and included a series of discussion points about the welfare of sheep and beef cattle (the focus group and interview scripts are provided in the supplementary materials). Questions were open-ended to allow participants to discuss their own thinking and ideas, and use their preferred concepts and language, rather than restricting their responses by providing a series of predetermined answers from which to select, as is typical with survey-based methods (Creswell and Creswell 2018). Asking the same questions in each focus group and interview provided a foundation from which to compare results. The research was considered complete when thematic saturation was reached based on iterative analysis rather than predetermined measures (Charmaz 2006). Focus groups and interviews were digitally recorded and fully transcribed, with each transcript compared to handwritten notes to check for accuracy and then anonymised. Each transcript was treated as a rich, narrative text where the first author inductively coded all responses in NVivo (Richards 2005) using methods similar to open coding (Strauss and Corbin 1990). However, our methods are not strictly based in traditional grounded theory (Corbin and Strauss 1990); instead, we used the generic inductive qualitative model (Maxwell 2005; Hood 2007) which blends the processes of description and interpretation during the generation of research questions, as well as purposeful sampling including demographic-based recruitment to strengthen our abilities to generalise cross-population and to other locations. Due to the qualitative nature of these research methods, we did not aim to generate strict representativeness or statistical significance (Hood 2007). Instead, we placed greater emphasis on what was said by participants rather than on how many participants made particular claims.1 Quotes used in the following results section are illustrative and typical of those coded to the theme developed during analysis. Results Overall, 66 meat consumers from the selected Australian locales participated in this research during 2015–16. Of these, 67% were women with ages ranging from 18 to 24 years to over 65. 50% were in Adelaide, South Australia; 31% percent were in Melbourne, Victoria; and 19% in Toowoomba, Queensland. We note that results did not differ across the three locales as perhaps might be expected elsewhere: unlike in some other countries, even people whom reside in larger cities are likely to have current or past contact with rural locales or past direct experience of agriculture based on migration history and basic geography. The overwhelming majority of participants self-identified as being of Anglo-Celtic descent, with only 4 participants (6%) identifying as being of an alternative ethnicity in our study. The dominance of Anglo-Celtic participants is important to note to contextualise our results and discussion, but is unsurprising given the relatively high percentages of Australians with this background, this study’s reliance on volunteers, and the inclusion of a regional centre amongst our study sites where non-Anglo-Celtics are much more uncommon.2 “That’s What I Expect, Anyway” Each focus group or interview began by asking participants to describe the images that came to mind when they thought of beef and sheep meat farming. Subsequently they were asked to identify things that could negatively impact animal welfare or if they had particular concerns (one prominent theme, animal transport, has been considered in a previous paper, Buddle et al. 2018b, so we do not explore it in detail here). Participants’ descriptions of animals on farms, their ideas about animal diets and animal behaviours, and their ideas about what was not normal for animals, and hence what impacted on animal welfare, revealed that allowing animals to behave ‘naturally’ was considered ‘good’ by participants. When asked to visualise sheep and beef cattle farming, most participants described animals “grazing in a paddock”, revealing that participants “expected” extensive3 production systems. Cows in the paddock. That’s all I see. Sheep in the paddock. (Sally, female, 45–54 years, Adelaide) Well basically sheep and cattle grazing in green farms with lots of room and space to walk around and enjoy themselves basically. That is what I expect anyway. (Tim, male, 55–64 years, Melbourne) Diet and living environment were intrinsically linked for many participants, who highlighted the importance of grazing as a natural behaviour. The ‘goodness’ of extensive production systems was contrasted with more intensive systems that restrict animal movement and the abilities of animals to express natural behaviours. I guess [sheep and cattle] are grazing animals. We know them as grazing animals. And grazing means walking around, eating little bits of stuff here and there so…If it can do its natural behaviour, yeah. Doing what is natural for an animal. (Henry, male, 35–44 years, Melbourne) I think, you know … an animal should be able to enjoy the things it naturally does. For a pig is to wallow and do its thing and that and for a chicken to be able to scratch and fluff its feathers, … [for] cattle to be crammed into those veal crates and of course the piggeries even, it isn’t good. (Tilly, female, 45–54 years, Adelaide) Participants also associated grain-feeding with animal confinement, emphasising a lack of choice for animals with respect to diet, and issues with animals being limited to foods that they would not ordinarily choose themselves. I don’t like grain-fed. I feel like, at least if they are grass-fed they are out in the pasture, they’re not just in a little shed somewhere, locked in a cage being fed rubbish they wouldn’t normally eat. So I avoid grain-fed. (Sarah, female, 35–44 years, Adelaide) When describing sheep and cattle production, the participants in this study described extensive production systems as ‘natural,’ ‘normal,’ and ‘expected’ in contrast to more intensive production systems where animal movement and dietary choice may be restricted. These findings align with previous research including opposition to animal confinement (see e.g. Miele and Evans 2005; Lassen et al. 2006; Sørensen and Fraser 2010; Boogaard et al. 2011b; Miele et al. 2011; Spooner et al. 2014; Coleman et al. 2016); the importance to consumers of animals expressing their ‘natural behaviours’ (Te Velde et al. 2002; Vanhonacker and Verbeke 2009; Vanhonacker et al. 2012; Spooner et al. 2014; Coleman et al. 2016), and preferences for grazing as part of good animal welfare (Verbeke et al. 2010; Spooner et al. 2014; Coleman et al. 2016). However, the expectation of extensive production methods for sheep and cattle in the Australian context is worth noting, given the frequent use of feedlots as mentioned in the introduction. “Australian Family Farms Care for Their Animals” The ‘good’ production systems that our participants described were connected with farms that were smaller in scale and family-owned, and represented what participants believed farming ‘should be’: I suppose I have had my grandfather’s example of an old school farmer and I remember him talking with disgust about another farmer who overstocked … He was disgusted with that because he’s…old style, and he looked after his land. (Tilly, female, 45–54 years, Adelaide) What is good farm animal welfare? And what is bad? Do you have a sense of that? How do you evaluate that yourself? Every animal’s got a name [focus group laughter]. But some of them do do that, like not every animal maybe, but some of their favourite ones. I’m just being silly. (Susie, female, 45–54 years, Adelaide) But I think that’s a good point because, yes treating each and every animal as an individual, living creature, that deserves a certain quality of life for the time that it’s alive and quality of death as well. Even if it doesn’t have a name, but I think that’s sort of the idea that you’re [implying]. (Chrissie, female, 35–44 years, Adelaide) In contrast, larger farms were associated with commercial activity, and the replacement of farmers with managers: I know there is [sic] a lot of family farms that really do care about their animals, but then you hear stories … about shearing where they just, like they’re cutting open sheep and they just like stitching them up without any kind of medication or anything and obviously the farmers must be aware of that and, so, I wonder if the commercial divide makes a difference…If the farmer really owns the farm and looks after the farm, or if the farmer’s just a manager, maybe. (Susie, female, 45–54 years, Adelaide) Participants’ responses suggest that the relationship between humans and animals is an important part of ensuring good animal welfare. Higher levels of care were attributed to owner-operators who were viewed as involved in every aspect of raising the animal, as compared to employees or those involved during only one stage of production. This point is also raised by Wilkie (2010) who suggests that the scale and type of animal production can significantly impact the extent to which producers can realistically engage with their animals. Presenting the relationship between farmer and animal as one of intimacy and care also is crucial to the construction of products as ethical and humanely produced (Heath and Meneley 2010). In a similar vein, Staples and Klein (2017) describe the alleged intimacy between humans and animals on smaller, ‘more ethical’ farms, in contrast to the perceived (and possibly exaggerated, see Baker 2013) separation between animals and humans in intensive farming. Participants also emphasised that farms owned and managed by Australians have better standards of animal welfare than those run by foreign companies and farmers. They claimed that Australian farmers care for the welfare of their animals and ‘others’ (often not explicitly defined) do not: …a lot of farmers, especially Australian ones, kind of look out for the welfare of animals (Lucy, female, 18–24 years, Toowoomba) Can you think of any practices that aren’t necessarily good for the animal’s welfare? Not really, no. I think Australia [sic] are pretty good, umm. Better not talk about the overseas ones though. (Mark, male, 45–54 years, Adelaide) So when you’re buying Australian, is that, because it’s local or also because we have good standards when it comes to animal management? It’s numerous things. It’s helping our own farmers, ah it’s helping the country. Plus I don’t trust [pause] Asian food. I’m really anti that. From a food hygiene perspective? Oh everything. Everything. I just don’t think their standards are up to anywhere near our standards, both in killing, producing …I’d be racist if I went any further. (Iain, male, over 65 years, Melbourne) “Happy not Knowing” It was notable that many participants avoided the topic of slaughter until it was directly introduced and tended not to focus on details about slaughter practices related to the meat that they consume. Participants indicated that they had limited knowledge about slaughter and would rather leave it that way: I think I am happy not knowing because I know what it was like on the farm and I mean I can just generally think what it would be like. I mean it is a trauma to the animal, it is a trauma to the people who look after the animal. (Susan, female, 55–64 years, Melbourne) However, participants noted that slaughter was necessary for meat production. When you think about sheep and beef cows and their welfare; do you ever associate the slaughter process? Yeah, you do, hey. You do. That’s the hard part to answer. It’s really, really hard cause we don’t [sic] to see it. No one [does]. And then people, I don’t even know how they can stand doing it. How can they work in that environment? A place like that. You know what? It’s got to be done. Unfortunately, it has to. How are we going to survive, to eat? I’m pretty sure you like eating your meat…and I know you, you’re the same way as me, we all feel for our animals, but the job’s got to be done. (Paddy, male, 35–44 years, Melbourne) Participants were willing to consume meat despite not wanting to consider or know about slaughter, but also recognised the trauma that animals likely face during the process: It’s quite cruel because the animal is a living thing ah they have life and sense and feeling but we are just slowly killing them it’s just like…eww. (Tristan, male, 18–24 years, Melbourne). “It should be Done Quickly” Despite not wanting to know about the details of the slaughter process, participants unanimously stated that slaughter of livestock should be performed as quickly and humanely as possible, echoing previous research from Europe (Miele and Evans 2005) and Canada (Spooner et al. 2014).[Slaughter] should be done quickly. Get it over and done with for the poor animals, you know so they’re not suffering. (Paddy, male, 35–44 years, Melbourne). Many participants suggested that animals not only had a right to a good life, but also a “good death: There are issues around having a good life and serving a good death. Or as good as possible. (Marian, female, 55–65 years, Melbourne) I think they should just be allowed to be animals right up until the day we have to kill them, you know what I mean? So they have a nice life. Like you, you need to honour that someone’s, that you, you are taking something’s life, we should give it a nice life right up until the time you have to take it. (Sarah, female, 35–44 years, Adelaide) Some participants also described the impact that they believed stress to have not only on animal welfare, but also meat quality. Although stress is known to have a direct negative effect on meat quality (Meat and Livestock Australia n.d.), participants expressed the connection in more general terms. A lot of time they are forced to take the animals long distances which often means putting them into, jamming them into trucks and putting the animals under an enormous amount of stress which I think is bad for the animals, it is bad for the meat or the meat quality. (Marian, female, 55–64 years, Melbourne). Most participants considered the practice of stunning prior to slaughter to be an essential part of a ‘good death’ for livestock animals: How any person can stand there and like, kill a live animal is beyond me. And as mum says halal meat is even worse. When they what, slit their throat and let them bleed to death. So it, it’s different between the two processes. However, you think both are cruel? Oh very. Yeah. One is kinder than the other though. At least one has been hit with a stun gun, before you know, they’re killed, as such. They’re not aware of it. Yeah which is kinder… halal is just disgusting. And how anyone can sit there and eat a piece of halal meat, knowing that this animal went through an enormous amount of pain is beyond me. I wouldn’t eat it. (Olivia, female, 25–34 years, Toowoomba) Many participants were concerned about animal treatment in destination countries that receive live exports from Australia, and particularly halal slaughter, which was often described as “disgusting.” Such reactions were used as justifications for banning live export, as participants considered Australia to have better slaughter practices than the predominantly Muslim destination countries. As an alternative to live export, participants often suggested that animals should be slaughtered in Australia and exported as prepared or processed meat, and considered this option to be economically viable and as creating jobs for Australians while removing the cruelty that they associated with the live export industry: I feel that the live animal export is a major issue as far as I am concerned personally and everyone in my circles. I think we can do far more for this country by bringing onshore the abattoirs and processing situations and if the Muslim countries insist on having our live animals then I am afraid it should be a closed door and we seek other markets that do accept our processed meat. (Joyce, female, over 65 years, Adelaide) Discussion In terms of animal welfare, all participants in this study had clear ideas about what makes a ‘good life’ and a ‘good death.’ Although they do not want to engage with the details about slaughter (likely for psychological reasons as discussed below), they have deep concerns about halal slaughter and related practices as being inhumane and cruel. Most participants associated animals doing and eating what is ‘natural’ with higher animal welfare, and linked such conditions to ‘expected’ and ‘normal’ ideas of livestock production. Many placed high value on Australian family farms and Australian farmers, in part because they believed that animals receive better treatment than on farms run by non-Australians. Although these themes may appear to be distinct from one another, we contend that they are embedded in long-standing cultural understandings and expectations of what Australian agriculture, particularly livestock production, ‘should be’ according to Anglo-Celtic Australians (the dominant background of our participants) and are associated with processes of ‘othering’ those of non-Anglo-Celtic background and Muslims in particular. In this study, we targeted meat consumers, which permitted us to assess what they wanted to know about the processes associated with meat production. It was striking that while most did not want to consider or think about slaughter, they recognised the trauma that animals likely face during the process. This kind of affected ignorance, “generated by what one knows but does not want to hear” (Schwartz 2018, p. 1), involves the refusal to think about or consider a practice, particularly if the practice may be immoral (Williams 2008). In addition, many of our participants’ responses reflect the “meat paradox,” which refers to the cognitive dissonance that arises for many between the recognition that consumption behaviours harm animals, and the enjoyment that results from eating meat as part of their diets (Herzog 2010; Joy 2010; Piazza et al. 2015; Dowsett et al. 2018). Even with increasing levels of ambivalence and moral aversion towards killing animals for a variety of reasons, and a rise in the number of people adopting plant-based diets (de Padilha et al. 2022), many in modern societies are still fond of consuming meat (Leroy and Praet 2017). Loughnan et al. (2014) note that cognitive dissonance can be resolved either by rejecting meat consumption entirely or through various psychological manoeuvres, both of which bring moral beliefs and attitudes into alignment with behaviours. Meat-eaters have been noted to use four justificatory categories to ease discomfort with the slaughter process, known as the 4Ns: meat is natural (part of human biological nature); normal (how we ought to behave based on our customs and traditions); necessary (nutrients found only in meat are essential for a healthy, balanced diet), and nice (people enjoy consuming meat) (Joy 2010; Piazza et al. 2015). While discussing slaughter as a necessary condition for eating meat, many participants in the current study explicitly emphasised various aspects of these principles of justification, for instance associating the consumption of meat with a healthy diet. Despite its prominence in Australian culture, agriculture currently only employs approximately 1.3% of Australia’s population, contributing about 3% to the country’s economic output (National Farmers’ Federation 2020),4 figures that are significantly diminished from less than a century ago when agriculture was the largest economic sector (Davidson and Brodie 2005). Although most Australians over the past two centuries have lived in large urban centres, Australia continues to be imagined through the “frontier narrative” (as noted earlier by Furniss 1999), including an Australian variant of the rural idyll known as “countrymindedness” (Aitkin 1985; Cockfield and Botterill 2012; see also Bell 2006), which portrays the family farm as “hard-working settlers making an honest living off the land” (Cairns et al. 2015, p. 1189). Stereotypical depictions of the idyllic family farm with an Anglo-Celtic white male in a flannelette shirt and Akubra hat have been used in popular TV shows (Phillipov 2017, ch. 3; Phillipov and Loyer 2019) and many marketing campaigns in recent years (Chen 2016b; Phillipov 2016). For example, Coles supermarket’s seminal “Helping Australia Grow” campaign used family farm images to help “connect Coles’ methods of sourcing fresh produce to the qualities of embeddedness, trust and place [usually] associated with alternative food practices” (Phillipov 2016, p. 590). Chen (2016b) argues that such depictions of farmers “reinforce the moral status quo…thanks to popular ignorance about the realities of animal use and farm life” (p. 122) and obscure the “reality that farms in Australia today comprise large agribusiness enterprises as well as family owned and operated properties” (p. 115). Nonetheless, many current advertising campaigns still echo these romanticised and idyllic themes about Australian agriculture. In many of the participants’ comments in this study, there was focus on concepts such as ‘expected’ and ‘normal’ in relation to the extensive production systems that are thought to be ‘good’ (and ‘natural’) for beef and sheep meat production in Australia. The comments describing expectations associated with extensive production systems draw heavily on the romanticised stereotypes noted above, and are in tension with the reality of the frequent use of feedlots in this sector as noted in the introduction. Importantly, these characterisations are distinct from popular understandings of agriculture in many other locales where the terminology used often refers to the ‘natural’ (such as in Sweden, see Saltzman et al. 2011). The types of ‘traditional’ values associated with the ‘natural’ in relation to farming in Europe (Verbeke et al. 2010; Boogaard et al. 2011a,c) are not present amongst this study’s Australian participants, but instead a distinct sense of what is ‘normal’ emerges here to describe what is ‘good’: that which is expected, typical, and acceptable. What is not ‘good’ in the processes associated with death, according to our participants, is clear: halal slaughter. This focus raises a number of issues. Firstly, in Australia, all animals must be stunned prior to slaughter, unless permission is granted for ritual slaughter (Australia and New Zealand Food Regulation Ministerial Council 2007, 7.12[1]). These exceptions compose a very small portion of the overall number of slaughtered animals, and primarily are provided for Kosher production; most importantly, Islamic religious leaders in Australia have accepted pre-slaughter stunning practices as halal (Bergauld-Blackler 2016; Armanios and Ergene 2018; Loyer et al. 2020).5 Despite the ubiquity of pre-stunning slaughter in Australia, including in halal slaughter, participants described halal practices as not including pre-slaughter stunning and hence causing a ‘bad’ death.6 On the surface, these concerns echo previous research indicating that Europeans (Vanhonacker et al. 2010) and Canadians (Spooner et al.2014) also consider pre-slaughter stunning to be important for animal welfare. They also are based in part on lack of awareness about slaughter processes in general, which relate to participants’ preferences to not consider these processes in any detail and the lack of regulatory transparency in this domain (see Loyer et al. 2020 for a review). In addition, concerns about halal slaughter are clearly associated with concerns about the live export trade (see Buddle et al. 2018b for more detail on Australians’ perceptions of livestock transport including in conjunction with live export), which was an issue that received considerable media attention in the few years prior to this research (Buddle and Bray 2019). Problems with welfare practices in the context of live export of livestock to Australia have been supported by graphic and troubling exposes, such as in the ABC TV Four Corners program, “A Bloody Business” (Doyle 2011), as well as frequent campaigns by animal welfare organisations (e.g., RSPCA n.d.). Live export was not a focus of the current study.7 However, no matter what our views may be on live export practices, it is important to note that Fozdar and Spittles (2014) argue that representation of mistreatment of Australian cattle in the program drew on deeply embedded Orientalist traditions of seeing so-called Western behaviours as civilised, while viewing Eastern practices as barbaric. What is of primary interest for our interpretation is that the concerns expressed in this research by many participants about halal slaughter in Australia appear to be associated with broader concerns about Islamic practices (see Senate 2015) and bias against non-Anglo-Celtic migrants as ‘other.’ Several of the quotes provide clear examples of ‘othering’, which occurs when there is a difference in beliefs or behaviours between two social or cultural groups, particularly when it is perceived that there are potential risks associated with the ‘other’ group. Most striking perhaps are the comments that emphasise participants’ visceral reactions to halal slaughter which in effect describe some practices associated more generally with slaughter in Australia (e.g., exsanguination or bleeding out which occurs in all forms of slaughter), but which they use to draw a distinction between halal and ‘standard’ slaughter practices. It is clear from many of our participants’ comments that concerns about halal slaughter are not associated with accurate knowledge or direct experiences. However, these comments do reflect deeper trends in current Australian culture, and the weaponisation of halal as part of an anti-Islamic movement (Wong and Millie 2015), particularly in the use of emotive language and the association of certain practices to people themselves in participants’ responses. There is significant stigma in Australia associated with the Islamic faith (40% of respondents in a major survey had a negative attitude towards Muslims when asked anonymously: see Markus 2019).8 Discrimination against those of colour and different races extends far back in Australian history: the informal measures and formal regulations that collectively are known as the “White Australia policy” effectively limited non-British migration to Australia from 1901 till the post-World War II period when the policy relaxed to allow refugees from continental Europe; this policy was not formally eliminated until the mid-1970s (Tavan 2005). Even after the enormous growth in migration and settlement of Muslims in Australia over the past 30 years, Islam is still viewed by many as culturally incompatible with the so-called Western ideals on which Australia was founded (Mansouri 2020). The increased numbers of arrivals of asylum seekers and refugees9 in the 1990s onward (Kabir 2005) and the recent “Operation Sovereign Borders” campaign by the Australian Government (2018), along with ongoing fears of Islamic extremist terror attacks (Fahd 2017), have contributed to discrimination against Muslims. As seen in the participants’ responses quoted above, such concerns and fears shape views on animal welfare, and particularly what makes for a good animal life and death. Consumer anxieties about food frequently involve processes of ‘othering’ (Jackson 2010), including what has been termed “culinary xenophobia,” or fear and distrust of foreign food (Santich 1996, 232; Edwards et al. 2000; Anderson and Benbow 2015). Our findings expand on this scholarship to explicitly include food-related behaviours including production practices. Such practices can bring people together, but also can set people apart, performing exclusionary roles that can lead to culinary xenophobia and reinforce ‘otherness’ (Wright and Annes 2013). Culinary xenophobia provides opportunities for identity formation and a way of establishing unity in the face of difficulty or disaster, particularly by distributing or shifting the blame to separate ‘us’ from the risky other (Milne 2013). We contend that our participants perceive that an animal’s ‘good death,’ and in turn the potential for a ‘good life’, are what is at risk here. Providing animals with good lives by allowing them to live what was termed ‘naturally’ (i.e., normally and in line with participants’ expectations) and a good death by minimising stress, were key concerns of many participants, and were linked to meat quality and even safety. The factors associated with animals living naturally were tied to a traditional, rural idyll, which as noted above is largely a symbolic landscape, although it is closely associated with national identity in this Australian context. Thus we contend that the values that many of our participants (who were overwhelmingly Anglo-Celtic) held about providing farm animals with good lives and good deaths are deeply entangled with these ideas of identity and what it means to be Australian. These values are expressed through the reflections of a rural idyll in expressions of what is ‘normal’ for sheep meat and beef production in Australia, but also in relation to the rejection of that which is variously viewed as ‘other,’ including intensive production, foreign ownership, and halal slaughter. Acknowledgements None Declarations Conflict of Interest On behalf of all authors, the corresponding author states that there is no conflict of interest. 1 We do use terms such as ‘most’ (a majority), ‘many’ (more than 50% but not a majority), ‘some’ (less than 50% but more than 25%), and so on in a common-sense manner, in the spirit of what Becker (1970) traditionally called ‘quasi statistics’; for a useful overview of debates associated with this issue, see Maxwell (2010). 2 According to combined relevant categories from the 2021 Australian Bureau of Statistics’ Australian Standard Classification of Cultural and Ethnic Groups, at least 51.7% of the Australian population is of Anglo-Celtic ancestry; this percentage is declining (down from as high as 75% as recently as 1987, see Khoo et al. 2003; p. 165) due to immigration from other locales, but remains an important cultural influence in Australia. 3 Throughout this paper, ‘extensive’ production systems refer to those where animals are raised in outdoor environments and housed within paddocks such as free-range and grass-fed production systems, while ‘intensive’ production systems refer to those in which animals are raised indoors and housed within barns or sheds as is common in conventional pork and poultry production, or in penned yards as is common with feedlots, with large quantities of inputs. 4 It is unclear whether these statistics include or exclude transport associated with agriculture, but this gap in available information does not lessen the difference between popular perceptions of the importance of this domain and its actual employment numbers, and the decline over the course of the 20th century. 5 Practices differ in other countries in Europe and elsewhere, although developments in meat production and processing have resulted in some aspects of ritualistic killing and associated ceremonial approaches to slaughter (including not pre-stunning) becoming irrelevant (Hoogland et al. 2005; Bergauld-Blackler et al. 2016; Leroy and Praet 2017). 6 It may well be that participants have seen media coverage of halal slaughter in overseas countries that do not stun animals and they presume that all halal slaughter occurs in the same way, in part based on their experience of other religions where there is less local interpretation of prescriptions such as the rules associated with halal slaughter. However investigation of this conjecture is beyond the scope of this paper; for further discussion on the differences in halal slaughter practices, see Bergauld-Blackler et al.2016). 7 As it is not a main focus in this paper, we do not provide details of the cases for and against live export, although there is voluminous evidence about best and problematic practices and their effects on animal welfare, as well as rogue operators. 8 We do not use more recent surveys as they focused heavily on COVID-19 related issues. 9 Asylum seekers are people seeking international protection but whose claims for refugee status have not yet been determined, whereas official refugees have had claims heard and approved. Although in the popular imagination the number of boat arrivals in particular is large in Australia, the number of people arriving unauthorised by boat is typically small in comparison to the numbers arriving elsewhere such as Europe. Overall, the number of asylum claims lodged in Australia is small in comparison to the USA and Europe, in part because of the difficulties in accessing Australia due to distance and geography. Nonetheless, migration remains a hotly contested issue, frequently covered in the popular media. Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References Aitkin D Countrymindedness: The Spread of an Idea Australian Cultural History 1985 4 34 41 Anderson L Benbow HM Cultural Indigestion in Multicultural Australia: Fear of ‘Foreign’ Foods in Australian Media Gastronomica: The Journal of Critical Food Studies 2015 15 1 34 43 10.1525/gfc.2015.15.1.34 Ankeny, R. A. 2008. The Moral Economy of Red Meat in Australia. In S.R. Friedland (ed.), Proceedings of the Oxford Symposium on Food and Cookery 2007 (pp. 20–28). Blackawton, Totnes: Prospect Books. Ankeny RA Pilcher J Food and ethical consumption The Oxford Handbook of Food History 2012 New York Oxford University Press 461 480 Armanios F Ergene B Halal Food: A History 2018 New York Oxford University Press Australia and New Zealand Food Regulation Ministerial Council Australian Standard for Hygienic Production and Transportation of Meat and Meat Products for Human Consumption 2007 Collingwood, VIC CSIRO Publishing Australian Bureau of Agricultural and Resource Economics and Sciences. 2016. Agricultural Commodities Statistics Report https://www.data.gov.au/data/dataset/pb_agcstd9abcc0022016_sn9dg, last accessed 2 September 2022. Australian Government (Department of Home Affairs). 2018. Operation Sovereign Borders (OSB) https://osb.homeaffairs.gov.au/, last accessed 2 September 2022. Australian Government (Department of Infrastructure and Regional Development). 2015. State of Australian Cities 2014–2015: Progress in Australian Regions https://www.infrastructure.gov.au/sites/default/files/migrated/infrastructure/pab/soac/files/2015_SoAC_full_report.pdf, last accessed 2 September 2022. Baker K Abbots E Lavis A Home and Heart, Hand and Eye: Unseen Links between Pigmen and Pigs in Industrial Farming Why We Eat, How We Eat: Contemporary Encounters Between Foods and Bodies 2013 Farnham, UK Ashgate 53 73 Becker, H. S. 1970. Field Work Evidence. In Sociological Work: Method and Substance (pp. 39–62), 1st edition. New Brunswick, NJ: Transaction Books. Bell D Cloke P Marsden T Mooney P Variations on the Rural Idyll The Handbook of Rural Studies 2006 London Sage 149 160 Bergauld-Blackler F Bergauld-Blackler F Fischer J Lever J The Halal Certification Market in Europe and the World: A First Panorama Halal Matters: Islam, Politics, and Markets in Global Perspective 2016 London Routledge 105 126 Bergauld-Blackler F Fischer J Lever J Halal Matters: Islam, Politics, and Markets in Global Perspective 2016 London Routledge Boogaard BK Bock BB Oosting SJ Wiskerke JSC Van der Zijpp AJ Social Acceptance of Dairy Farming: The Ambivalence between the Two Faces of Modernity Journal of Agricultural and Environmental Ethics 2011 24 3 259 282 10.1007/s10806-010-9256-4 Boogaard BK Boekhorst LJS Oosting SJ Sorensen JT Socio-cultural Sustainability in Pig Production: Citizen Perspectives in the Netherlands and Denmark Livestock Science 2011 140 1–3 189 200 10.1016/j.livsci.2011.03.028 Boogaard BK Oosting SJ Bock BB Wiskerke JSC The Sociocultural Sustainability of Livestock Farming: An Inquiry into Social Perceptions of Dairy Farming Animals 2011 5 9 1458 1466 Bray HJ Ankeny RA ‘Happier chickens lay tastier eggs’: Motivations for Buying Free-Range Eggs in Australia Anthrozoös 2017 30 2 213 226 10.1080/08927936.2017.1310986 Broom DM Sentience and Animal Welfare 2014 Wallingford, UK CABI Buddle EA Bray HJ How Farm Animal Welfare Issues are Framed in the Australian Media Journal of Agricultural and Environmental Ethics 2019 32 3 357 376 10.1007/s10806-019-09778-z Buddle EA Bray HJ Pitchford WS Keeping it ‘inside the fence’: An Examination of Responses to a Farm Animal Welfare Issue on Twitter Animal Production Science 2017 58 3 435 444 10.1071/AN16634 Buddle EA Bray HJ Ankeny RA ‘Why would we believe them?’ Meat Consumers’ Reactions to Online Farm Animal Welfare Activism in Australia Communication Research and Practice 2018 4 3 246 260 10.1080/22041451.2018.1451209 Buddle, E.A., Bray, H.J., and Ankeny R.A. (2018b). “I feel sorry for them”: Australian meat consumers’ perceptions about sheep and beef cattle transportation. Animals, 8(10), 10.3390/ani8100171 Bush AJ Hair JF An Assessment of Mall Intercept as a Data Collection Method Journal of Marketing Research 1985 22 2 158 167 10.1177/002224378502200205 Cairns K McPhail D Chevrier C Bucklaschuk J The Family Behind the Farm: Race and Affective Geographies of Manitoba Pork Production Antipode 2015 47 5 1184 1202 10.1111/anti.12147 Carey R Parker C Scrinis G Capturing the Meaning of ‘Free Range’: The Contest between Producers, Supermarkets and Consumers for the Higher Welfare Egg Label in Australia Journal of Rural Studies 2017 54 266 275 10.1016/j.jrurstud.2017.06.014 Charmaz K Constructing Grounded Theory 2006 Thousand Oaks, CA Sage Chen P Animal Welfare in Australia: Politics and Policy 2016 Sydney Sydney University Press Chen P In the Media Animal Welfare in Australia: Politics and Policy 2016 Sydney Sydney University Press 85 122 Cockfield G Botterill LC Signs of Countrymindedness: A Survey of Attitudes to Rural Industries and People Australian Journal of Political Science 2012 47 4 609 622 10.1080/10361146.2012.731482 Coleman GJ Jongman E Greenfield L Hemsworth P Farmer and Public Attitudes toward Lamb Finishing Systems Journal of Applied Animal Welfare Science 2016 19 2 198 209 10.1080/10888705.2015.1127766 26882113 Corbin J Strauss A Grounded Theory Research: Procedures, Canons, and Evaluative Criteria Qualitative Sociology 1990 13 1 3 21 10.1007/BF00988593 Creswell JW Qualitative Inquiry and Research Design: Choosing among Five Approaches 2013 3 Thousand Oaks, CA Sage Creswell JW Creswell JD Research Design: Qualitative, Quantitative and Mixed Method Approaches 2018 Thousand Oaks, CA Sage Davidson G Brodie M Struggle Country: The Rural Ideal in Twentieth Century Australia 2005 Sydney Monash University ePress de Oliveira Padilha, L. G. Malek, L. and W. Umberger. 2022. Consumers’ attitudes towards lab-grown meat, conventionally raised meat and plant-based protein alternatives. Food Quality and Preference, 99, 10.1016/j.foodqual.2022.104573. Denzin NK Lincoln YA The Sage Handbook of Qualitative Research 2011 4 Thousand Oaks, CA Sage Dixon J Authority, Power and Value in Contemporary Industrial Food Systems International Journal of Sociology Agriculture and Food 2003 11 1 31 39 Dowsett, E., C. Semmler, H. J. Bray, R. A. Ankeny, and A. Chur-Hansen. 2018. Neutralising the Meat Paradox: Cognitive Dissonance, Gender and Eating Animals. Appetite 123, 280–288. Doyle, M. (Producer). 2011. A Bloody Business. In S. Spencer (Executive Producer), Four Corners Australian Broadcasting Corporation. Edwards L Occhipinti S Ryan S Food and Immigration: The Indigestion Trope Contests the Sophistication Narrative Journal of Intercultural Studies 2000 21 3 297 308 10.1080/07256860020007458 Fahd, C. 2017. Terror, Muslims, and a Culture of Fear: Challenging the Media Messages. The Conversation https://theconversation.com/terror-muslims-and-a-culture-of-fear-challenging-the-media-messages-77170, last accessed 2 September 2022. Fozdar F Spittles B Of Cows and Men: Nationalism and Australian Cow Making The Australian Journal of Anthropology 2014 25 1 73 90 10.1111/taja.12067 Furniss E The Burden of History: Colonialism and the Frontier Myth in a Rural Canadian Community 1999 Vancouver University of British Colombia Press Future Eye. 2018. Commodity or Sentient Being – Australia’s Shifting Mindest on Farm Animal Welfare https://www.sheepcentral.com/wp-content/uploads/2019/05/190129-Commodity-or-Sentient-Being-Australias-Shifting-Mindset-on-Farm-Animal-Welfare-v.-7.0.pdf, last accessed 2 September 2022. Heath D Meneley A The Naturecultures of Foie Gras Food Culture and Society 2010 13 3 421 452 10.2752/175174410X12699432701024 Herzog H Some We Love, Some We Hate, Some We Eat: Why It’s So Hard to Think Straight about Animals 2010 New York Harper Perennial Hood JC Bryant A Charmaz K Orthodoxy vs. Power: The Defining Traits of Grounded Theory The SAGE Handbook of Grounded Theory 2007 London SAGE 151 164 Hoogland CT de Boer J Boersema JJ Transparency of the Meat Chain in the Light of Food Culture and History Appetite 2005 45 1 15 23 10.1016/j.appet.2005.01.010 15949870 Jackson P Food Stories: Consumption in an Age of Anxiety Cultural Geographies 2010 17 2 147 165 10.1177/1474474010363844 Joy M Why We Love Dogs, Eat Pigs and Wear Cows: An Introduction to Carnism 2010 San Francisco, CA Red Wheel/Weiser Kabir N Muslims in Australia 2005 Sydney UNSW Press Khoo S-A McDonald P Khoo S-E The Transformation of Australia’s Population: 1970–2030 2003 Sydney UNSW Press Lassen J Sandøe P Forkamn B Happy Pigs are Dirty! Conflicting Perspectives of Animal Welfare Livestock Science 2006 103 3 221 230 10.1016/j.livsci.2006.05.008 Leroy F Praet I Animal Killing and Postdomestic Meat Production Journal of Agricultural and Environmental Ethics 2017 30 67 86 10.1007/s10806-017-9654-y Lewis T Huber A A Revolution in an Eggcup?: Supermarket Wars, Celebrity Chefs and Ethical Consumption Food Culture & Society 2015 18 2 289 307 10.2752/175174415X14190821960798 Loughnan S Bastian B Haslam N The Psychology of Eating Animals Current Directions in Psychological Science 2014 23 2 104 108 10.1177/0963721414525781 Loyer J Whittaker AL Buddle EA Ankeny RA A Review of Legal Regulation of Religious Slaughter in Australia: Failure to Regulate or a Regulatory Fail? Animals 2020 10 9 1530 10.3390/ani10091530 32872627 Malek L Umberger WJ Goddard E Committed vs. Uncommitted Meat Eaters: Understanding Willingness to Change Protein Consumption Appetite 2019 138 115 126 10.1016/j.appet.2019.03.024 30917940 Malhotra N Marketing Research: An Applied Orientation 2006 3 Sydney Pearson/Prentice Hall Mansouri F Islam and Muslims in Australia: The Social Experiences of Early Settlement and the Politics of Contemporary Race Relations Politics and Religion Journal 2020 14 1 10.54561/prj1401127m María GA Public Perception of Farm Animal Welfare in Spain Livestock Science 2006 103 3 250 256 10.1016/j.livsci.2006.05.011 Markus, A. 2019. Mapping Social Cohesion: The Scanlon Foundation Surveys 2019. https://scanlonfoundation.org.au/archived-research/mapping-social-cohesion-survey-2019/, last accessed 2 September 2022. Maxwell JA Qualitative Research Design: An Interactive Approach 2005 Thousand Oaks, CA Sage Maxwell JA Using Numbers in Qualitative Research Qualitative Inquiry 2010 16 6 475 482 10.1177/1077800410364740 Meat and Livestock Australia. 2018. Industry Overview. https://www.mla.com.au/about-mla/Cattle-sheep-goat-industries/industry-overview/. Meat and Livestock Australia. (n.d.). Factors affecting eating quality. https://www.mla.com.au/research-and-development/meat-standards-australia/factors-affecting-eating-quality/, last accessed 2 September 2022. Michelletti M Southerton D Political and ethical consumption Encyclopedia of Consumer Culture, Part III 2011 London Sage 1097 1100 Miele, M., and A. Evans. 2005. European Consumers’ Views about Farm Animal Welfare Welfare Quality Conference, Science and Society Improving Animal Welfare, Brussels, Belgium. Miele M Veissier I Evans A Botreau R Animal Welfare: Establishing a Dialogue between Science and Society Animal Welfare 2011 20 103 117 Milne R Jackson P Group CONANX Risk Food Words: Essays in Culinary Culture 2013 London Bloomsbury 173 181 Miranda-de la Lama GC Estévez-Moreno LX Sepúlveda WS Estrada-Chavero MC Rayas-Amor AA Villarroel M María GA Mexican Consumers’ Perceptions and Attitudes towards Farm Animal Welfare and Willingness to Pay for Welfare Friendly Meats Meat Science 2017 125 106 113 10.1016/j.meatsci.2016.12.001 27940228 Muringai V Goddard E Bruce H Plastow G Ma L Trust and Consumer Preference for Pig Production Attributes in Canada Canadian Journal of Agricultural Economics 2017 65 3 477 514 10.1111/cjag.12138 National Farmers’ Federation. 2020. 2030 Roadmap: Australian Agriculture’s Plan for a $100 Billion Industryhttps://nff.org.au/wp-content/uploads/2020/02/NFF_Roadmap_2030_FINAL.pdf, last accessed 2 September 2022. National Health and Medical Research Council. 2007. updated 2018). National Statement on Ethical Conduct in Human Research https://www.nhmrc.gov.au/guidelines-publications/e72, last accessed 2 September 2022. Phillipov M ‘Helping Australia Grow’: Supermarkets, Television Cooking Shows, and the Strategic Manufacture of Consumer Trust Agriculture and Human Values 2016 33 3 587 596 10.1007/s10460-015-9643-6 Phillipov M Media and Food Industries: The New Politics of Food 2017 London Palgrave Macmillian Phillipov, M., and J. Loyer. 2019. In the Wake of the Supermarket ‘Milk Wars’: Media, Farmers and the Power of Pastoral Sentimentality. Discourse Context & Media 32, 10.1016/j.dcm.2019.100346. Piazza, J., M. B. Ruby, S. Loughnan, M. Luong, J. Kulik, H. M. Watkins, and M. Seigerman. 2015. Rationalizing Meat Consumption: The 4Ns. Appetite 91, 114–128. Richards L Handling Qualitative Data: A Practical Guide 2005 London Sage Publications Rodan D Mummery J The ‘Make It Possible’ Multimedia Campaign: Generating a New ‘Everyday’ in Animal Welfare Media International Australia 2014 153 1 78 87 10.1177/1329878X1415300110 RSPCA. (n.d.). Slaughter without Stunning. https://www.rspca.org.au/take-action/slaughter-without-stunning, last accessed 2 September 2022. Saltzman K Head L Stenseke M Do Cows Belong in Nature? The Cultural Basis of Agriculture in Sweden and Australia Journal of Rural Studies 2011 27 1 54 62 10.1016/j.jrurstud.2010.09.001 Salvin HE Lees AM Cafe LM Colditz IG Lee C Welfare of beef cattle in Australian feedlots: a review of the risks and measures Animal Production Science 2020 60 1569 1590 10.1071/AN19621 Santich B What the Doctors Ordered: 150 Years of Dietary Advice in Australia 1995 South Melbourne Hyland House Publishing Pty Ltd Santich B Looking for Flavour 1996 Adelaide Wakefield Press Santich B Cottle D Kahn L Beef Consumption: Historical Overview, Recent Trends and Contemporary Attitudes Beef Cattle Production and Trade 2014 Clayton South, VIC CSIRO Publishing 1 16 Schwartz B The Animal Welfare Battle: The Production of Affected Ignorance in the Swedish Meat Industry Debate Culture and Organization 2018 10.1080/14759551.2018.1513937 The Senate (Economics References Committee) Third Party Certification of Food 2015 Canberra Senate Printing Unit, Parliament of Australia Sinclair M Derkley T Fryer C Phillips CJC Australian Public Opinions Regarding the Live Export Trade Before and After an Animal Welfare Expose Animals 2018 8 7 106 10.3390/ani8070106 29966300 Sørensen JT Fraser D On-farm Welfare Assessment for Regulatory Purposes: Issues and Possible Solutions Livestock Science 2010 131 1 1 7 10.1016/j.livsci.2010.02.025 Spooner J Schuppli C Fraser D Attitudes of Canadian Citizens toward Farm Animal Welfare: A Qualitative Study Livestock Science 2014 163 1 150 158 10.1016/j.livsci.2014.02.011 Staples J Klein JA Consumer and Consumed Ethnos 2017 82 2 193 212 10.1080/00141844.2015.1107604 Strauss AL Corbin JM Social Sciences: Statistical Methods; Grounded Theory; Qualitative Research 1990 Newbury Park, CA Sage Tavan G The Long, Slow Death of White Australia 2005 Carlton North, VIC Scribe Te Velde H Aarts N Van Woerkum C Dealing with Ambivalence: Farmers’ and Consumers’ Perceptions of Animal Welfare in Livestock Breeding Journal of Agricultural and Environmental Ethics 2002 15 2 203 219 10.1023/A:1015012403331 Vanhonacker F Verbeke W Buying Higher Welfare Poultry Products? Profiling Flemish Consumers Who Do and Do Not Poultry Science 2009 88 12 2702 2711 10.3382/ps.2009-00259 Vanhonacker F Van Poucke E Tuyttens FAM Verbeke W Citizens’ Views on Farm Animal Welfare and Related Information Provision: Exploratory Insights from Flanders, Belgium Journal of Agricultural and Environmental Ethics 2010 23 6 551 569 10.1007/s10806-010-9235-9 Vanhonacker F Verbeke W Van Poucke E Pieniak Z Nijs G Tuyttens FAM The Concept of Farm Animal Welfare: Citizen Perceptions and Stakeholder Opinions in Flanders, Belgium Journal of Agricultural and Environmental Ethics 2012 25 79 101 10.1007/s10806-010-9299-6 Van Pouke E Vanhonacker F Griet N Braeckman J Verbeke W Tuyttens FAM Kaiser M Lien ME Defining the Concept of Animal Welfare: Integrating the Opinion of Citizens and Other Stakeholders Ethics and the Politics of Food 2006 Wageningen, Netherlands Wageningen Academic Publishers 555 559 Vargas-Bello-Pérez E Miranda-de la Lama GC Teixeria D Enríquez-Hidalgo D Tadich T Lensick J Farm Animal Welfare influences on Market and Consumer Attitudes in Latin America: The Cases of Mexico, Chile and Brazil Journal of Agricultural and Environmental Ethics 2017 30 5 697 713 10.1007/s10806-017-9695-2 Verbeke W Perez-Cueto FJA de Barcellos MD Krystallis A Grunert KG European Citizen and Consumer Attitudes and Preferences Regarding Beef and Pork Meat Science 2010 84 2 284 292 10.1016/j.meatsci.2009.05.001 20374787 Wilkerson A Judging, Tasting, Knowing “Good” Food Food Culture & Society 2016 19 2 223 226 10.1080/15528014.2016.1178524 Wilkie R Livestock/Deadstock: Working with Farm aAnimals from Birth to Slaughter 2010 Philadelphia, PA Temple University Press Williams NM Affected Ignorance and Animal Suffering: Why Our Failure to Debate Factory Farming Puts Us at Moral Risk Journal of Agricultural and Environmental Ethics 2008 21 4 371 384 10.1007/s10806-008-9087-8 Wong, J., and J. Millie. 2015. Explainer: What is Halal and How Does Certification Work? The Conversation, https://theconversation.com/explainer-what-is-halal-and-how-does certification-work-36300, last accessed 2 September 2022. Wright W Annes A Halal on the Menu? Contested Food Politics and French Identity in Fast Food Journal of Rural Studies 2013 32 388 399 10.1016/j.jrurstud.2013.08.001

0

PMC9753876

NO-CC CODE

2022-12-16 23:26:20

no

Food Ethics. 2023 Dec 15; 8(1):5

utf-8

Food Ethics

10.1007/s41055-022-00114-2

oa_other

==== Front Evol Psychol Sci Evol Psychol Sci Evolutionary Psychological Science 2198-9885 Springer International Publishing Cham 348 10.1007/s40806-022-00348-7 Research Article Responses to COVID-19 Threats: an Evolutionary Psychological Analysis Colarelli Stephen M. colar1sm@cmich.edu 1 Mirando Tyler J. t.mirando125@gmail.com 1 Han Kyunghee han1k@cmich.edu 1 Li Norman P. normanli@smu.edu.sg 2 Vespi Carter vespi1c@cmich.edu 1 Klein Katherine A. klein1k@cmich.edu 1 Fales Charles P. fales1cp@cmich.edu 1 1 grid.253856.f 0000 0001 2113 4110 Department of Psychology, Central Michigan University, Mount Pleasant, USA 2 grid.412634.6 0000 0001 0697 8112 School of Social Science, Singapore Management University, Singapore, Singapore 15 12 2022 111 27 6 2022 8 11 2022 8 11 2022 © The Author(s), under exclusive licence to Springer Nature Switzerland AG 2022, Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. Responses to COVID-19 public health interventions have been lukewarm. For example, only 64% of the US population has received at least two vaccinations. Because most public health interventions require people to behave in ways that are evolutionarily novel, evolutionary psychological theory and research on mismatch theory, the behavioral immune system, and individual differences can help us gain a better understanding of how people respond to public health information. Primary sources of threat information during the pandemic (particularly in early phases) were geographic differences in morbidity and mortality statistics. We argue that people are unlikely to respond to this type of evolutionarily novel information, particularly under conditions of high uncertainty. However, because individual differences affect threat perceptions, some individual differences will be associated with threat responses. We conducted two studies (during Phase 1 and 2 years later), using data from primarily public sources. We found that state-level COVID-19 morbidity and mortality rates had no relationship with mental health symptoms (an early indicator of how people were responding to the pandemic), suggesting that people—in general—were not attending to this type of information. This result is consistent with the evolutionary psychological explanation that statistical information is likely to have a weak effect on the behavioral immune system. We also found that individual differences (neuroticism, IQ, age, and political ideology) affected how people responded to COVID-19 threats, supporting a niche-picking explanation. We conclude with suggestions for future research and suggestions for improving interventions and promoting greater compliance. Keywords COVID-19 Mismatch Decision-making Individual differences Behavioral immune system ==== Body pmcThe number of deaths in the USA from the COVID-19 pandemic (over 1 million) has exceeded the US total from the Spanish influenza epidemic (Curley, 2021), which had been the deadliest pandemic in the US history (Barry, 2020).1 Worldwide, over 6 million people have died from COVID-19. While there have been significant improvements in scientific and public understanding of the disease, progress with public health interventions remains disappointing (Ishak, 2022; Lewis, 2021; Nan et al., 2022). For example, despite the severity of the COVID-19 pandemic and the widespread availability of safe and effective vaccines (the best-known way to defeat the pandemic), only 64% of the US population was fully vaccinated at the beginning of 2022 (Center for Disease Control and Prevention, 2021). Throughout the world, only 59% of the population has been vaccinated with at least two doses (Ritchie et al., 2020). These rates are much lower than for other serious infectious diseases—approximately 83% for polio, DPT, and measles (Muhoza et al., 2021). Although the COVID-19 pandemic will likely run its course, there will be other pandemics (Olshaker & Osterholm, 2017). Moreover, given the unprecedented scale of global travel, future pandemics have the potential to be as widespread and severe as the COVID-19 pandemic. While much has been achieved, challenges remain for improving how public health information is presented and how people respond to it (Ash et al., 2022). Guided by theory and research in three areas of evolutionary psychology—mismatch, the behavioral immune system, and individual differences—we undertook the present study to gain a better understanding of how people respond to COVID-19 threat information under different conditions as well as to examine psychological mechanisms that influence those responses. Primary sources of COVID-19 threat information (particularly during Phase 1 of the pandemic) were morbidity and mortality statistics (by geography) as well as non-personalized media images. We argue that people are unlikely to respond to this type of evolutionarily novel information, particularly under conditions of high uncertainty. However, because individual differences affect threat perceptions, some individual differences will be associated with threat responses. We examined people’s responses to the pandemic in the USA at two points in time—during Phase 1 and 2 years later. In Study 1, we examined the relationship between variation in state-level COVID-19 threat information and mental health symptoms (as indications of concern about COVID threats), as well as the relationship between personality characteristics and mental health symptoms. In Study 2, conducted 2 years later, we used available (state-level) data to examine the effects of intelligence, age, and ideology on vaccination rates. Timeline of the COVID-19 Pandemic During the first phase of the pandemic in the USA (March–June 2020), connections among the virus (SARS-CoV-2), infection, symptoms, and morbidity were causally opaque (Koelle et al., 2022). It was initially unknown who was most susceptible to infection, what groups would suffer the worst effects, and when during the course of infection the virus was most transmittable (Slifka & Gao, 2020; World Health Organization, 2020). Messages about wearing masks varied from unnecessary to a good idea to essential (Eikenberry et al., 2020; Worby & Chang, 2020). What was known was that morbidity and mortality rates varied by geography (CDC COVID-19 Response Team, 2020). Public health interventions were limited to keeping the public informed of morbidity and mortality rates and imposing lockdowns to prevent the spread of the virus. Thus, many types of direct metrics (e.g., vaccination, masking, and social distancing rates) that could be used to assess people’s threat responses were not available during Phase 1. A number of researchers argued that mental health symptoms were one of the few available indicators of how people were responding to the pandemic and to public health information—assuming that greater concern about or fear of the virus should manifest in greater emotional distress (Cullen et al., 2020; McGinty et al., 2020; Pfefferbaum & North, 2020). Two years later, the situation was considerably different. Although most areas of the USA experienced spikes in morbidity and mortality, lockdowns had mostly ended. There was a better scientific understanding of the virus and disease. Studies found that children were least at risk (Pierce et al., 2022), the elderly were most vulnerable (Liu et al., 2020), the primary vectors were respiratory droplets and contact routes, and infected people were most likely to transmit the virus when they were pre-symptomatic (Johansson et al., 2021). The evidence was clearer that masking and social distancing helped to minimize the chances of infection. Most important, effective vaccines became widely available, which provided a direct metric for assessing how people were responding to COVID-19 threats. COVID-19, Mismatch, and the Behavioral Immune System A likely assumption underlying the reporting of morbidity and mortality rates during Phase 1 was that people would respond “rationally” (e.g., Cushman, 2020; Eiser et al., 2012) to this information.2 That is, they would attend to the information, weigh the pros and cons of alternative courses of action, and behave according to threat levels. Where threat levels were high, people would behave more cautiously, be more concerned about health risks, and report greater levels of distress and anxiety. Conversely, where local threat levels were low, people would be less cautious, less concerned about becoming infected, and less anxious. Would this be the case? An evolutionary psychological perspective would suggest the opposite: people, in general, would not respond differentially to statistical information, particularly under conditions of high uncertainty (Colarelli & Thompson, 2008; Moore, 1996). Over millennia, our behavioral immune system has been the primary way that people responded to infectious disease threats—by detecting and avoiding pathogens (Schaller & Duncan, 2007). It operates by triggering avoidance responses to animate and inanimate objects that—recurrently over our evolutionary history—had a high probability of carrying infectious pathogens. This system triggers an avoidance response through the emotion of disgust (Cepon-Robins et al., 2021; Oaten et al., 2009). Both animate and inanimate objects with obvious signs of carrying infectious pathogens trigger the disgust response. Examples include spoiled food, feces, cadavers, sick animals, and people with noticeable signs of illness (e.g., blemishes, pustules, vomiting, a runny nose, skin pallor, deformities). The disgust response is normally followed by avoidance. More proactive (cultural) responses can also develop, including adherence to social norms and rituals that help guard against infection (e.g., personal hygiene, food preparation customs). However, because the behavioral immune system evolved in ancestral environments that are quite different from modern environments (Li et al., 2018), it is unlikely that people—in general—will respond to evolutionarily novel information, such as pandemic statistics and commentaries about global trends.3 However, as we discuss later, some individual differences are likely to influence how people respond to COVID-19 threat information. One might counter that the media and social media were full of images related to the ravages of COVID-19—and that these images should activate the behavioral immune system. While there were abundant images of overworked health care workers, intensive care units overflowing with COVID-19 patients, spikey balls representing the virus, and even coffins, there have been—for a variety of reasons, including medical privacy laws—few images of identifiable victims in the throes of the disease (Lewis, 2020). Moreover, images of intensive care units filled with COVID-19 patients, ventilators, and so forth are evolutionarily novel. Graphic images of sick and suffering people would most likely activate the behavioral immune system (Schaller et al., 2010). This was the case during the polio epidemics of the mid-twentieth century, with widespread public health campaigns using images of crippled children (see Fig. 1) (Mayo Clinic Staff, 2022).4 In addition, people infected with the coronavirus are initially asymptomatic, and early symptoms are not severe (resembling the common cold). By the time people were severely ill, many were out of view—isolated—at home or in a hospital. As a result, there were few inputs to activate the evolved behavioral immune system. Not surprisingly, it was people with direct experience with COVID-19—either experiencing symptoms themselves or having a friend or relative who was sick with COVID-19—that were most likely to suffer negative mental health symptoms (González-Sanguino et al., 2020).Fig. 1 Typical images of victims of the polio epidemic and COVID-19 pandemic We expand behavioral immune system research to investigate the role of evolutionarily novel stimuli associated with infectious diseases. Research on the behavioral immune system has traditionally examined how exposure to pathogenic stimuli—stimuli that, over evolutionary history, have recurrently been associated with infectious disease—creates a disgust response (or perceived vulnerability to disease), and how this in turn motivates avoidance behavior. We examine whether evolutionarily novel stimuli (specifically, variation in disease threat exposure statistics) creates differential vulnerability responses. We expect that the behavioral immune system may be less responsive to evolutionarily novel threat stimuli (Schaller et al., 2021).5 Individual Differences and Threat Information In large measure, evolutionary psychology focuses on broad, species-level adaptations and behavior. For example, the behavioral immune system is a species-typical mechanism that evolved to detect infectious pathogens and motivate people to avoid them. Yet, individual differences inevitably create variation in species-typical responses, including how people respond to disease (and other) threat information. While it is important to understand how people in general respond to infectious disease threats, it is also critical to understand how individual differences affect responses. This helps in building public health policies that can be tailored to individuals who may respond outside of population norms. Recent research in evolutionary psychology provides useful frameworks for thinking about how individual differences influence threat responses. Theories of niche picking, reactive heritability, and frequency-dependent selection suggest how individual differences in personality can evolve and be adaptive in different circumstances. Neuroticism continues to be widespread because people with this trait, over evolutionary history, were more likely to survive and reproduce by playing it safe. Given that hypervigilance to threat information can be an adaptive response for people with neurotic personalities, these individuals would be most likely to suffer depression, loneliness, or anxiety in the face of imminent danger, which in turn would trigger caution and isolating behavior. In contrast, other traits (e.g., extraversion, openness to experience) continue to be widespread because of the adaptive value of pursuing risky strategies. Ignoring threat information and carrying on normally may provide access to valuable opportunities, despite the risks (Nettle, 2006). Several studies conducted in the early phases of the pandemic, for example, found that neuroticism (Airaksinen et al., 2021) and perceived vulnerability to infection (Makhanova & Shepherd, 2020) stimulated protective responses. Thus, we expect that neuroticism would be associated with indices of emotional distress, while this is unlikely to be the case for other Big Five personality traits. The savanna-IQ interaction hypothesis suggests how general intelligence evolved to allow individuals to overcome problems associated with a mismatch (Kanazawa, 2010). That is, differences in intelligence should affect how people assess and act upon evolutionarily novel information (to the extent that causal connections are not entirely opaque). Thus, as the relationship between vaccines and disease mitigation became clearer, we would expect a positive relationship between intelligence and vaccination rates. In addition, evolution designed a variety of psychological adaptations that can be switched on and off throughout the lifespan (Buss, 2009). One of which is that people become more risk averse as they age (Rolison et al., 2014). As people age, they become more frail and susceptible to injury and infection.6 Thus, we would expect that older people would become more responsive to COVID-19 threats—to the extent that relatively clear and believable threat information was available. Finally, beliefs and ideology are memes that, given sufficient time, can evolve into norms and cultural practices on which people differ (Campbell, 1975; Henrich, 2016). Through cultural evolution, beliefs can become protective. However, they can also be unreliable (or harmful) over the short-term (Henrich, 2016)—such as some beliefs about COVID-19 threats and responses to threats (Conway et al., 2021; van Holm et al., 2020). In particular, anti-science beliefs and associated norms have, in some cases, become more of a signal of in-group-out-group membership (Boykin et al., 2021) than helpful responses to minimize COVID-19 infection. Study 1 During Phase 1 of the pandemic, people were exposed to two primary types of threat information that are evolutionarily novel: statistical reports of morbidity and mortality rates across different geographic regions and non-personalized images related to the virus, its treatment, and its effects. In addition, by the time people presented with serious symptoms, they were often isolated. We would therefore expect that people, on average, would not respond to variations on COVID-19 threat levels during Phase 1. That is, people in areas with high rates of mortality and morbidity should exhibit similar levels of mental distress as those in areas with lower mortality and morbidity rates. However, personality traits affect how people respond to threat information. In particular, traits such as neuroticism are associated with a higher threat sensitivity level (Barlow et al., 2014) and should be strongly associated with emotional distress. Method In Study 1, 418 individuals (67% response rate) across the USA were recruited during the third week of May 2020 using Amazon’s Mechanical Turk. However, only 291 participants representing 13 states (see Table 1) were included after removing states with less than 10 participants. The mean age of the sample was 37.76 years (SD = 10.98 years). Of the 291 participants, 54.0% were male, 45.7% were female, and 0.30% identified as other. The self-reported major racial/ethnic composition of the sample was 68.4% Caucasian, 18.2% Black, 10.0% Asian, 6.5% Hispanic/Latino, 3.4% Native American, 0.7% Middle Eastern, 0.7% Hawaiian, and 0.3% Other. Most of the participants were married (49.1%), while 27.5% were single, 16.5% were in a relationship/not married, 5.5% were divorced/separated, and 1.4% were widowed. Education categories for the sample were as follows: high school degree or equivalent (6.5%), some college but no degree (11.7%), associate degree (11.0%), bachelor’s degree (52.9%), and graduate degree (17.9%).Table 1 Sample size, mean (standard deviation) of loneliness, anxiety, and depression by state State n Loneliness Anxiety Depression M (SD) M (SD) M (SD) California 55 2.39 (0.88) 1.86 (0.72) 1.72 (0.79) Florida 29 2.28 (0.87) 1.76 (0.64) 1.53 (0.65) Georgia 19 2.39 (0.53) 1.75 (0.67) 1.43 (0.68) Illinois 18 2.47 (0.91) 1.93 (0.80) 1.77 (0.73) Michigan 10 2.20 (0.67) 1.77 (0.95) 1.39 (0.49) New York 32 2.54 (0.85) 1.93 (0.71) 1.65 (0.81) North Carolina 20 2.21 (1.06) 1.88 (0.93) 1.63 (0.80) Ohio 17 2.65 (0.94) 1.81 (0.51) 1.67 (0.72) Pennsylvania 16 2.65 (0.98) 2.09 (0.66) 1.78 (0.72) Texas 40 2.40 (0.76) 1.72 (0.68) 1.63 (0.74) Virginia 15 1.99 (0.70) 1.37 (0.46) 1.44 (0.67) Washington 10 2.66 (0.45) 2.03 (0.64) 1.84 (0.78) Wisconsin 10 2.76 (0.55) 1.98 (0.59) 1.86 (0.82) Loneliness (1–5 scale), anxiety (1–4 scale), depression (1–4 scale) We assessed state-level COVID-19 indicators with archival data from the COVID-19 Tracking Project at The Atlantic (The COVID-19 Tracking Project at The Atlantic, 2020). These data were gathered from state health websites across the nation. State-level infection and mortality rates were selected and divided by the population of each state. We also captured variations in length of government-imposed lockdowns between states using data from USA Today (2020). Data provided by USA Today were based on official reports of state-level lockdowns and aggregated by Safe Graph (a California-based firm). For mental health indicators, we measured loneliness with The Loneliness Scale (De Jong Gierveld & Van Tilburg, 1999), an 11-item Likert-style questionnaire (α = 0.83), depression severity with the 9-item Patient Health Questionnaire (α = 0.94; Kroenke & Spitzer, 2002), and state-anxiety with the 5-item shortened version of The State-Trait Inventory for Cognitive and Somatic Anxiety (α = 0.88; Marteau & Bekker, 1992). We measured the personality traits of neuroticism (α = 0.73), extraversion (α = 0.79), conscientiousness (α = 0.70), openness (α = 0.79), and agreeableness (α = 0.76), with the Mini-IPIP (Donnellan et al., 2006). Results and Discussion Table 1 presents the means and standard deviations of mental health variables (loneliness, anxiety, and depression) for 13 states. The lowest and highest means across states for each mental health variable were as follows: 1.99 (Virginia) to Wisconsin (2.76) for loneliness, 1.37 (Virginia) to Pennsylvania (2.09) for anxiety, and 1.39 (Michigan) to Wisconsin (1.86) for depression. Table 2 presents correlations among state-level COVID-19 threats (number of days locked down, infection rate, and mortality rate) and mental health symptoms (loneliness, anxiety, and depression). The number of days locked down was not significantly related to any of the outcome variables (r =  − 0.01 to 0.03). The mental health symptoms were not significantly related to either infection (r = 0.00 to 0.06) or mortality rates (r = − 0.00 to 0.07).Table 2 Correlations among state-level COVID-19 threats, mental health outcomes, and personality Loneliness Anxiety Depression State-level indicators of COVID-19a Number of days locked down − .01 − .00    .03 Infection rate    .06    .06    .00 Mortality rate    .06    .07 − .00 Personality traitsb Neuroticism    .57    .59    .57 Extraversion − .18 − .15 − .12 Openness − .29 − .23 − .24 Agreeableness − .36 − .23 − .26 Conscientiousness − .41 − .39 − .42 an = 291; p > .05 for all rs bn = 418; p < .05 for all rs Figure 2 shows mean scores of loneliness, anxiety, and depression across 13 states. States are listed in an ascending order of days of lockdown. Georgia had the shortest lockdown length (27 days), whereas Michigan had the longest lockdown (73 days). We expected an upward monotonic trend if days of lockdown were positively related to loneliness. However, no linear pattern was observed, suggesting that loneliness mean scores were not associated with the length of lockdowns. Similar results were found for infection and mortality rates. Although mean mental health outcome scores did not vary in a meaningful way across states, personality traits were associated with loneliness, depression, and anxiety (see Table 2) (r =  − 0.42 to 0.59, p < 0.05).Fig. 2 Mean mental health outcome scores by days of lockdown, infection rate, and mortality rate Our findings suggesting no association between threats of COVID-19 and mental health symptoms are similar to findings in other studies examining objective measures of COVID-19 threats and mental health (Nocentini et al., 2021). As we argued, people may not be attending to or believe implications of morbidity and mortality rates from COVID-19. However, our results also suggest that some people may be more sensitive to real or imagined health risks. We found that people low in emotional stability (i.e., high in neuroticism) reported greater negative mental health symptoms, while those with higher levels of extraversion, agreeableness, conscientiousness, and openness to experience reported greater levels of psychological well-being. Study 2 In Study 2, we examined how intelligence, age, and political ideology affected threat responses (vaccination rates) to COVID-19 2 years after the initial outbreak. Although information about the virus remained evolutionarily novel (morbidity and mortality statistics, non-personalized images), causal connections were clearer, the nature of the virus was better understood, and vaccines were widely available. Thus, we expected that intelligence and age would be positively related to vaccination rates, whereas an ideology that was suspicious of experts and science would be negatively related.7 Method Current total vaccination rates for each US state (as of February 6th, 2022) were gathered from the Mayo Clinic’s website (Mayo Clinic, 2022). The Mayo Clinic is a nonprofit hospital system and academic medical center that provides esteemed, often publicly-accessible medical research. We collected vaccination rates by age group (ages 5–11, 12–17, 18–64, and 65 +) for each US state from the Mayo Clinic website, and the average vaccination rate for each age group was calculated. We collected IQ data for each state from the World Population Review website (World Population Review, 2022). The World Population Review is an independent organization that seeks to provide normally inaccessible demographic data for public examination and use. For the data used in the present study, the World Population Review used a study conducted by the Washington Post that aggregated various measures of cognitive ability (IQ scores, SAT and ACT scores, and education level) into an overall IQ measure for each US state. Data pertaining to 2020 presidential election results were drawn from CNN’s website (CNN, 2020), which tracked which states were won by Joe Biden or Donald Trump, as well as the percentage of the votes going to either of the two candidates for each state. Results The mean full vaccination rate for each age group across all the US states is shown in Table 3. Vaccination rates increased with age, with the youngest age group (5–11) showing the lowest percent vaccinated (M = 21.89, SD = 9.60) and the oldest age group (65 +) showing the largest percent vaccinated (M = 94.08, SD = 4.65). The 65 + age group also showed the least amount of variability in vaccination (SD = 4.65) of all the age groups.Table 3 Percent of the US population fully vaccinated by age group Age group M SD 5–11 21.89 9.60 12–17 53.72 12.97 18–64 68.28 9.30 65 +  94.08 4.65 Total 59.71 27.81 Percentages were averaged across all the US states for each age group The relationship between the total vaccination rates of the US states and each state’s average IQ was tested, along with the relationship between states’ total vaccination rates and the total percentage of each state’s vote that went to Trump in the 2020 presidential election. There was a significant positive correlation between the percentage of the population fully vaccinated and average IQ across all states (r = 0.35, p < 0.001). There was also a significant and strong negative correlation between total vaccination rate and the percentage of the vote that went to Trump across all states (r = − 0.88, p < 0.001).8 These results also show that individual differences influence how people responded to COVID-19 pandemic information—in this case, even when vaccines and more information about the virus were available. Both intelligence and age correlated positively with vaccination rates, while political ideology (support for Donald Trump in the 2020 presidential election) correlated negatively with vaccination rates. General Discussion We found that state-level COVID-19 threat information (number of days locked down, infection rates, and mortality rates) had no relationship with mental health symptoms (loneliness, anxiety, and depression) during the early months of the pandemic. Mean scores of mental health variables did not vary in a meaningful way across states. Our finding of a lack of association between information about COVID-19 threat exposure and mental health is consistent with other studies examining objective measures of COVID-19 threats and mental health (Nocentini et al., 2021). These findings are consistent with evolutionary psychological explanations. The behavioral immune system is unlikely to be triggered by evolutionarily novel information, such as statistical information and non-personal images (Li et al., 2018). However, individual differences were associated with how people responded to COVID-19 threats. Even when causal information about the pandemic was vague (during Phase 1), neuroticism correlated strongly with negative mental health symptoms (loneliness, anxiety, and depression). All of the other traits correlated negatively with adverse mental health symptoms. While our data are cross sectional, these correlations suggest some impact of the pandemic: they are stronger than typical correlations among the Big Five and negative mental health outcomes prior to the pandemic (e.g., Bunevicius et al., 2008). Individual differences played an important role in how people responded to COVID-19 threats 2 years after the outbreak of the pandemic when causal connections were clearer and after vaccines were available. The elderly (65 +), who were most at risk from COVID-19, had considerably higher rates of being fully vaccinated (94%) than all other age groups. Aggregate state IQ levels correlated positively with aggregate vaccination rates. Thus, although the behavioral immune system is unlikely to respond to abstract information (such as infection and mortality statistics), perceived vulnerability and intelligence can, to some extent, counteract this. Limitations A limitation of Study 1 is that the mismatch implication—people not having evolved to respond to novel stimuli such as threat information—was supported by the absence rather than the presence of significant results. Moreover, the lack of correlations between COVID-19 threat indicators and mental health could be due to many unidentified factors. Thus, while we contribute to the literature by outlining a potentially strong explanation for a pressing, real-world phenomenon, we only provide indirect empirical support for the mismatch hypothesis. More rigorous tests—including experimental methods that manipulate different ways of conveying the virus (e.g., evolutionarily novel statistical reporting versus visual presentation of severe outcomes) are clearly needed to substantiate the hypotheses. Another promising route is to investigate moderators that may influence when the correlation between COVID-19 and mental health, as well as vaccination and other precautions, becomes significant. Another possible limitation is the use of state-level data. It would have been preferable if threat level data in Study 1 were at a smaller unit of analysis (e.g., county). In Study 2, it would have been preferable if our data were at the individual level. However, certain types of individual data relevant to our research questions (e.g., matching individual vaccination status with IQ, who an individual voted for) would be restricted and possibly inaccurate. Nevertheless, we believe that our results are broadly indicative of the relationships we assessed. We recommend that more granular research should be conducted in future studies to delve further into these relationships. Implications Despite the mitigating influence of scientific literacy and education, the remediation of modern, global pandemics through public health interventions is and will remain difficult. Most public health interventions provide evolutionarily novel information and require people to behave in ways that are mismatched with evolved human perceptual and decision-making mechanisms. This includes understanding and accepting abstract scientific information, avoiding or staying distant from people who do not seem ill, staying at home when feeling fine, wearing face coverings, and getting injected with foreign substances. The greater the degree that a desired behavior is at odds with its adaptive value over millennia of human evolution, the more difficult it will be for an intervention to effectively encourage that behavior (Jones, 2001). For example, because frequent social interaction with friends and family has been adaptive to humans for millennia, people will be more resistant to public health interventions that restrict normal human interaction (lockdowns, social distancing, wearing facemasks) than to interventions that facilitate social interaction. Our findings, combined with the above evolutionary logic, have four major implications for public policy. First, expecting broad voluntary compliance—especially during COVID-19-like pandemics—is unrealistic. For the majority of people, some mandatory regulations may be necessary to assure sufficient compliance, particularly during the early stages of an outbreak. This can occur through mandates from government or other institutions, such as employers. Typically, countries with stronger vaccine mandates and social pressure for vaccination have higher vaccination rates (Suliman et al., 2021). Interventions that link compliance with valued evolutionary-based rewards (such as status, access to status, or mating opportunities) are more likely to be successful. For example, making admission to workplaces, schools, and gathering places for singles contingent upon wearing masks or having proof of vaccination is likely to increase compliance. Second, because people respond selectively to pandemic threats based on individual differences, communication strategies should be selectively tailored to specific groups. People who are most likely to be affected by a pandemic—the elderly in the case of COVID-19 and parents of young children in the case of polio—are more likely to use effortful appraisal—what Kahneman (2011) calls System 2 thinking and what others have referred to as systematic or central processing (Petty & Cacioppo, 1986). Thus, information and appeals to the most vulnerable groups should be designed to engage more elaborate processing, such as the presentation of high-quality and accessible scientifically backed arguments. Groups that consider themselves to be less vulnerable—and thus are less motivated to carefully process information—may be persuaded by more superficial methods such as using attractive celebrity endorsements (Petty & Cacioppo, 1986). Third, when populations are facing disease-based threats, public health information should be tailored so that it activates the behavioral immune system. With COVID-19, more work should be done on examining appropriate and believable imagery in public health communications. Such public health interventions were successful polio vaccination and anti-smoking campaigns. Unfortunately, this approach has been little used COVID-19 public health campaigns as well as with some mismatched-based lifestyle diseases (obesity, Type II diabetes) where it would clearly be appropriate. Finally, it must be acknowledged that belief systems are difficult to change. If people believe that a vaccination is unsafe or that the negative effects of the disease are overstated, which is increasingly common due to media misinformation, it will be difficult to change those beliefs with a rational argument based on scientific evidence. People use reason to find justifications for their beliefs, which in turn enhance their reputations within specific groups—not to find the true state of affairs (Mercier & Sperber, 2017; Yong et al., 2021). Thus, to overcome opposition to public health policies, clearer explanations of their scientific basis are unlikely to be effective. A better strategy to get through to a skeptical public would be to use positive public health testimonials from high status individuals who are from those groups in which a majority of members are resistant to public health interventions. This would be a much more effective way of changing norms (Henrich, 2016).9 In summary, the COVID-19 pandemic has generated a less than desirable response in places where people are relatively free to choose that response. We have provided an explanation based on evolutionary psychological principles and obtained empirical results consistent with this explanation and inconsistent with a more commonly accepted explanation. More work is needed, but findings from a growing number of studies indicate that a consideration of how the modern world is mismatched to how we have evolved to think, feel, and behave can provide insights into the numerous problems that humans are now facing and why they are difficult to overcome. Author Contribution TJM, KAK, and CPF collected data for Study 1. CV collected data for Study2. KH conducted the statistical analyses. SMC and TJM wrote the initial drafts. SMC and NPL wrote the latter drafts. TJM and SMC initiated the study. KH, NPL, TJM, CV, and SMC edited later drafts. Funding This research did not receive funding from public, commercial, or not-for-profit sectors. Data availability Data are available from the first author upon request. Declarations Ethics approval Individual data were collected anonymously; all other data were from publically available archival sources. The study was therefore exempt from institutional review board approval (IRB#: 2020–173). Consent to participate Informed consent was obtained from all individual participants. Consent for publication All authors approved this submission. Conflict of interest The authors declare no competing interests. 1 Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) is the virus that causes COVID-19. 2 Providing statistical information on food labels and menus, particularly the calorie content of food items, has been a common public health intervention to attempt to curb the worldwide epidemic of obesity. It has also been unsuccessful (Kiszko et al., 2014). Folwarczny et al. (2022) argue that evolutionary psychological-based interventions would be more effective. 3 Several studies have found that people respond to statistical information about threats (disease, storms). Two studies found that they respond with over-perception bias (Makhanova et al., 2015; Miller & Maner, 2012). Another study (Bacon & Corr, 2020) found that perceived vulnerability to disease increased slightly after participants in an experiment read COVID-19 morbidity and mortality statistics. Because these were lab studies, it is likely that subjects’ attention was specifically focused COVID-19 statistics, which is less probable in a natural setting. Thus, research in a variety of settings is important for understanding how people assess risk to natural threats (Eiser et al., 2012) and how the outcomes found in lab studies relate to avoidance emotions and behavior in natural settings. 4 The polio epidemic was one of the most (if not the most) feared disease in the first half of the twentieth century. Unlike COVID-19, the symptoms of polio were obvious—paralysis and often death, primarily among infants and children (Baicus, 2012; Centers for Disease Control and Prevention (CDC), 2021). Prior to the vaccine and when the polio pandemic was at its worst (1948–1955), parents were extraordinarily cautious about letting their children go outside or to public gathering places (e.g., swimming pools), particularly in the summers, when the incidences of infection were highest (Mayo Clinic Staff, 2022). 5 People responded pretty much the same during the Spanish Flu pandemic on the early twentieth century as they did now—ignoring public health recommendations to social distance and wear masks (Barry, 2020). 6 With children, of course, vaccinations are parents’ decisions. We expect, though, that parents would follow a decision calculus based on the perceived threat of COVID-19 to their children’s well-being. As the evidence became clear that young children were least susceptible to infection and least likely to become ill or die from Covid (Pierce et al., 2022), we would expect that children would be the group least likely to be vaccinated. During the polio epidemic children were the most susceptible demographic group, and they were most likely to be vaccinated (Mayo Clinic Staff, 2022). 7 At the beginning of the pandemic, when causal connections were opaque and before effective vaccines were available, there was some evidence that the elderly were no more cautious than people of middle age (Daoust, 2020). 8 For elderly group, vaccination % ranged from 83.20 (Arkansas) to 99.90 (VT, RI, Main, WA, NH, MN, DE, WI) in 50 states. The mean % vaccination for states that voted for Biden was 96.67 (SD = 3.69), and for Trump, it was 91.48 (SD = 4.05). 9 Among the most successful public strategies in the anti-smoking campaigns were using graphic images of the ravages of smoking and invoking peer pressure from high status role models (e.g., Farrelly et al., 2012; Hurd et al., 1980). Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References Airaksinen J Komulainen K Jokela M Gluschkoff K Big Five personality traits and COVID-19 precautionary behaviors among older adults in Europe Aging and Health Research 2021 1 100038 10.1016/j.ahr.2021.100038 34568860 Ash C Alderton G Kelly P Scanlon ST Vinson V Wible B A time to reflect: Lessons from 2 years of the COVID-19 pandemic Science 2022 375 1098 1099 10.1126/science.abo7976 35271327 Bacon AM Corr PJ Behavioral immune system responses to coronavirus: A reinforcement sensitivity theory explanation of conformity, warmth toward others and attitudes toward lockdown Frontiers in Psychology 2020 11 566237 10.3389/fpsyg.2020.566237 33324279 Baicus A History of polio vaccination World Journal of Virology 2012 1 108 114 10.5501/wjv.v1.i4.108 24175215 Barlow, D. H., Sauer-Zavala, S., Carl, J. R., Bullis, J. R., & Ellard, K. K. (2014). The nature, diagnosis, and treatment of neuroticism: Back to the future. Clinical Psychological Science, 2, 344–365. 10.1177/2167702613505532 Barry, J. M. (2020). The great influenza: The story of the deadliest pandemic in history. Penguin UK. Boykin K Brown M Macchione AL Drea KM Sacco DF Noncompliance with masking as a coalitional signal to US conservatives in a pandemic Evolutionary Psychological Science 2021 7 3 232 238 10.1007/s40806-021-00277-x 33747752 Bunevicius A Katkute A Bunevicius R Symptoms of anxiety and depression in medical students and in humanities students: Relationship with big-five personality dimensions and vulnerability to stress International Journal of Social Psychiatry 2008 54 6 494 501 10.1177/0020764008090843 18974188 Buss DM How can evolutionary psychology successfully explain personality and individual differences? Perspectives on Psychological Science 2009 4 4 359 366 10.1111/j.1745-6924.2009.01138.x 26158983 Campbell DT On the conflicts between biological and social evolution and between psychology and moral tradition American Psychologist 1975 30 1103 1126 10.1037/0003-066X.30.12.1103 Center for Disease Control and Prevention. (2021, September 25). COVID-19 vaccinations in the United States. https://covid.cdc.gov/covid-data-tracker/#vaccinations_vacc-total-admin-rate-total CDC COVID-19 Response Team (2020). Geographic differences in COVID-19 cases, deaths, and incidence - United States, February 12-April 7, 2020. MMWR. Morbidity and Mortality Weekly Report, 69(15), 465–471. 10.15585/mmwr.mm6915e4 Cepon-Robins TJ Blackwell AD Gildner TE Liebert MA Urlacher SS Madimenos FC Sugiyama LS Pathogen disgust sensitivity protects against infection in a high pathogen environment Proceedings of the National Academy of Sciences 2021 118 e2018552118 10.1073/pnas.2018552118 Colarelli S Thompson M Stubborn reliance on human nature in employee selection: Statistical decision aids are evolutionarily novel Industrial and Organizational Psychology 2008 1 3 347 351 10.1111/j.1754-9434.2008.00060.x Conway LG Woodard SR Zubrod A Chan L Why are conservatives less concerned about the coronavirus (COVID-19) than liberals? Comparing political, experiential, and partisan messaging explanations Personality and Individual Differences 2021 183 1 10 10.1016/j.paid.2021.111124 CNN. (2020). Presidential Results. CNN Politics. https://www.cnn.com/election/2020/results/president Cullen, W., Gulati, G., & Kelly, B. D. (2020). Mental health in the COVID-19 pandemic. QJM: An International Journal of Medicine, 113(5), 311–312. Curley, C. (2021, September 22). How did this happen? The U. S. COVID-19 death toll has surpassed the 1918 flue total. Healthline. https://www.healthline.com/health-news/how-did-this-happen-the-u-s-covid-19-death-toll-has-surpassed-1918-flu-total Cushman F Rationalization is rational Behavioral and Brain Sciences 2020 43 e28 10.1017/S0140525X19001730 De Jong Gierveld J Van Tilburg TG Manual of the Loneliness Scale 1999 VU University Amsterdam Daoust J. F. (2020). Elderly people and responses to COVID-19 in 27 Countries. PLoS ONE (7): e0235590. 10.1371/journal.pone.0235590 Donnellan MB Oswald FL Baird BM Lucas RE The Mini-IPIP Scales: Tiny-yet-effective measures of the Big Five factors of personality Psychological Assessment 2006 18 2 1192 1203 10.1037/1040-3590.18.2.192 Eikenberry SE Mancuso M Iboi E Phan T Eikenberry K Kuang Y Gumel AB To mask or not to mask: Modeling the potential for face mask use by the general public to curtail the COVID-19 pandemic Infectious Disease Modelling 2020 5 293 308 10.1016/j.idm.2020.04.001 32355904 Eiser JR Bostrom A Burton I Johnston DM McClure J Paton D White MP Risk interpretation and action: A conceptual framework for responses to natural hazards International Journal of Disaster Risk Reduction 2012 1 5 16 10.1016/j.ijdrr.2012.05.002 Farrelly MC Duke JC Davis KC Nonnemaker JM Kamyab K Willett JG Juster HR Promotion of smoking cessation with emotional and/or graphic antismoking advertising American Journal of Preventive Medicine 2012 43 5 475 482 10.1016/j.amepre.2012.07.023 23079169 Folwarczny M Otterbring T Sigurdsson V Tan LK Li NP Old minds, new marketplaces: How evolved psychological mechanisms trigger mismatched food preferences Evolutionary Behavioral Sciences 2022 10.1037/ebs0000288 González-Sanguino C Ausín B Castellanos MÁ Saiz J López-Gómez A Ugidos C Muñoz M Mental health consequences during the initial stage of the 2020 Coronavirus pandemic (COVID-19) in Spain Brain, Behavior, and Immunity 2020 87 172 176 10.1016/j.bbi.2020.05.040 32405150 Henrich J The secret of our success 2016 Princeton University Press Hurd PD Johnson CA Pechacek T Bast LP Jacobs DR Luepker RV Prevention of cigarette smoking in seventh grade students Journal of Behavioral Medicine 1980 3 1 15 28 10.1007/BF00844911 7401172 Ishak, N. (2022, May 22). Stronger Public Health Systems and uniform messaging are critical to fight covid-19. Vox. Retrieved May 23, 2022, from https://www.vox.com/2022/5/22/23136724/covid-19-pandemic-public-health-response-summer-2022 Johansson MA Quandelacy TM Kada S Prasad PV Steele M Brooks JT Butler JC SARS-CoV-2 transmission from people without COVID-19 symptoms JAMA Network Open 2021 4 1 e2035057 e2035057 10.1001/jamanetworkopen.2020.35057 33410879 Jones OD Time-shifted rationality and the law of law’s leverage: Behavioral economics meets behavioral biology Northwestern University Law Review 2001 95 1141 1205 Kahneman D Thinking, fast and slow 2011 Macmillan Kanazawa S Evolutionary psychology and intelligence research American Psychologist 2010 65 1 279 289 10.1037/a0019378 20455621 Kiszko KM Martinez OD Abrams C Elbel B The influence of calorie labeling on food orders and consumption: A review of the literature Journal of Community Health 2014 39 6 1248 1269 10.1007/s10900-014-9876-0 24760208 Koelle K Martin MA Antia R Lopman B Dean NE The changing epidemiology of SARS-CoV-2 Science 2022 375 1116 1121 10.1126/science.abm4915 35271324 Kroenke K Spitzer RL The PHQ-9: A new depression diagnostic and severity measure Psychiatric Annals 2002 32 9 1 7 10.3928/0048-5713-20020901-06 Lewis M The premonition: A pandemic story 2021 Norton Lewis, S. E. (2020, May 1). Where are the photos of people dying of Covid? New York Times. https://www.nytimes.com/2020/05/01/opinion/coronavirus-photography.html Li NP van Vugt M Colarelli SM The evolutionary mismatch hypothesis: Implications for psychological science Current Directions in Psychological Science 2018 27 38 44 10.1177/0963721417731378 Liu K Chen Y Lin R Han K Clinical features of COVID-19 in elderly patients: A comparison with young and middle-aged patients Journal of Infection 2020 80 14 18 10.1016/j.jinf.2020.03.005 Makhanova A Miller SL Maner JK Germs and the out-group: Chronic and situational disease concerns affect intergroup categorization Evolutionary Behavioral Sciences 2015 9 8 19 10.1037/ebs0000028 Makhanova A Shepherd MA Behavioral immune system linked to responses to the threat of COVID-19 Personality and Individual Differences 2020 167 110221 10.1016/j.paid.2020.110221 32834281 Marteau, T. M. & Bekker, H. (1992). The development of a six-item short form of the State Scale of the Spielberger State-Trait Anxiety Inventory (STAI). British Journal of Clinical Psychology, 31(3), 301–306. 10.1111/j.2044-8260.1992.tb00997.x Mayo Clinic. (2022). U.S. COVID-19 vaccine tracker: See your state’s progress. https://www.mayoclinic.org/coronavirus-covid-19/vaccine-tracker Mayo Clinic Staff. (2022). History of Polio: Outbreaks and vaccine timeline. https://www.mayoclinic.org/coronavirus-covid-19/history-disease-outbreaks-vaccine-timeline/polio McGinty EE Presskresicher R Han H Psychological distress and loneliness reported by US adults in 2018 and April 2020 Journal of American Medical Association 2020 324 1 93 94 10.1001/jama.2020.9740 Mercier H Sperber D The enigma of reason 2017 Harvard University Press Miller SL Maner JK Overperceiving disease cues: The basic cognition of the behavioral immune system Journal of Personality and Social Psychology 2012 102 1198 1213 10.1037/a0027198 22329656 Moore RF Caring for identified versus statistical lives: An evolutionary view of medical distributive justice Ethology and Sociobiology 1996 17 6 379 401 10.1016/S0162-3095(96)00079-9 Muhoza, P., Danovaro-Holliday, M. C., Diallo, M. S., Murphy, P., Sodha, S. V., Requejo, J. H., & Wallace, A. S. (2021). Routine Vaccination Coverage—Worldwide, 2020. Morbidity and Mortality Weekly Report, 70(43), 1495. https://www.cdc.gov/mmwr/volumes/70/wr/mm7043a1.htm Nan X Iles IA Yang B Ma Z Public Health Messaging during the COVID-19 Pandemic and Beyond: Lessons from Communication Science Health Communication 2022 37 1 19 10.1080/10410236.2021.1994910 34724838 Nettle D The evolution of personality variation in humans and other animals American Psychologist 2006 20 622 631 10.1037/0003-066X.61.6.622 Nocentini, A., Palladino, B. E., & Mensini, E. (2021). Adolescents’ stress reactions in response to COVID-19 pandemic at the peak of the outbreak in Italy. Clinical Psychological Science, 9(3), 1–8. 10.1177/2167702621995761 Oaten M Stevenson RJ Case TI Disgust as a disease-avoidance mechanism Psychological Bulletin 2009 135 303 321 10.1037/a0014823 19254082 Olshaker, M., & Osterholm, M. T. (2017). Deadliest enemy: our war against killer germs. Hachette UK. Petty, R. E., & Cacioppo, J. T. (1986). Communication and persuasion: Central and peripheral routes to attitude change. New York: Springer-Verlag. Pfefferbaum B North CS Mental health and the COVID-19 pandemic New England Journal of Medicine 2020 383 6 510 512 10.1056/NEJMp2008017 32283003 Pierce CA Herold KC Herold BC Chou J Randolph A Kane B Hensley SE COVID-19 and children Science 2022 377 6611 1144 1149 10.1126/science.ade1675 36074833 Ritchie, H., Mathieu, E., Rodés-Guirao, L., Appel, C., Giattino, C., Ortiz-Ospina, E., Hasell, J., Macdonald, B., Beltekian D., & Roser, M., (2020). Coronavirus Pandemic (COVID-19). Published online at OurWorldInData.org. Retrieved from: 'https://ourworldindata.org/coronavirus' [Online Resource] Rolison JJ Hanoch Y Wood S Liu PJ Risk-taking differences across the adult life span: A question of age and domain Journals of Gerontology Series b: Psychological Sciences and Social Sciences 2014 69 6 870 880 10.1093/geronb/gbt081 24149517 Schaller M Duncan LA Forgas JP Haselton MG von Hippel W The behavioral immune system: Its evolution and social psychological implications Evolution and the social mind: Evolutionary psychology and social cognition 2007 Routledge/Taylor & Francis Group 293 307 Schaller M Miller GE Gervais WM Yager S Chen E Mere visual perception of other people’s disease symptoms facilitates a more aggressive immune response Psychological Science 2010 21 649 652 10.1177/0956797610368064 20483842 Schaller, M., Murray, D. R., & Hofer, M. K. (2021). The behavioural immune system and pandemic psychology: the evolved psychology of disease-avoidance and its implications for attitudes, behaviour, and public health during epidemic outbreaks. European Review of Social Psychology, 1–37. Slifka MK Gao L Is presymptomatic spread a major contributor to COVID-19 transmission? Nature Medicine 2020 26 10 1531 1533 10.1038/s41591-020-1046-6 Suliman DM Nawaz FA Mohanan P Modber MAKA Musa MK Musa MB Moonesar IA UAE efforts in promoting COVID-19 vaccination and building vaccine confidence Vaccine 2021 39 6341 6345 10.1016/j.vaccine.2021.09.015 34561138 The COVID Tracking Project at The Atlantic. (2020). Coronavirus numbers by state. https://docs.google.com/spreadsheets/u/2/d/e/2PACX-1vRwAqp96T9sYYq2-i7Tj0pvTf6XVHjDSMIKBdZHXiCGGdNC0ypEU9NbngS8mxea55JuCFuua1MUeOj5/pubhtml# USA Today. (2020). Coronavirus reopening map of COVID-19 case trends, reopening status and mobility. Retrieved October 11, 2020. https://www.usatoday.com/storytelling/coronavirus-reopening-america-map/ van Holm, E. J., Monaghan, J., Shahar, D. C., Messina, J. P., & Surprenant, C. (2020). The impact of political ideology on concern and behavior during COVID-19. Available at SSRN 3573224. Worby CJ Chang HH Face mask use in the general population and optimal resource allocation during the COVID-19 pandemic Nature Communications 2020 11 1 1 9 10.1038/s41467-020-17922-x World Health Organization. (2020). Coronavirus disease 2019 (COVID-19): Situation Report, 73. World Population Review (2022). Average IQ by State 2022. https://worldpopulationreview.com/state-rankings/average-iq-by-state Yong JC Li NP Kanazawa S Not so much rational but rationalizing: Humans evolved as coherence-seeking, fiction-making animals American Psychologist 2021 76 781 793 10.1037/amp0000674 33151700

0

PMC9753878

NO-CC CODE

2022-12-16 23:26:20

no

Evol Psychol Sci. 2022 Dec 15;:1-11

utf-8

Evol Psychol Sci

10.1007/s40806-022-00348-7

oa_other

==== Front SN Soc Sci SN Soc Sci Sn Social Sciences 2662-9283 Springer International Publishing Cham 570 10.1007/s43545-022-00570-x Original Paper Qualitative content analysis of Nigerian heads-of-state and presidents’ inaugural addresses: text mining, topic modelling and sentiment analysis Fowobaje Kayode Raphael 1 http://orcid.org/0000-0002-1312-944X Mashood Lawal Olumuyiwa maslaw008@gmail.com 2 Ekholuenetale Michael 1 http://orcid.org/0000-0003-4982-5209 Ibidoja Olayemi Joshua ojibidoja@fugusau.edu.ng 3 1 grid.9582.6 0000 0004 1794 5983 Department of Epidemiology and Medical Statistics, Faculty of Public Health, College of Medicine, University of Ibadan, Ibadan, Nigeria 2 Department of Statistics, Faculty of Science, Air Force Institute of Technology, Nigerian Air Force Base, Mando, Kaduna State, Nigeria 3 Department of Mathematics, Faculty of Science, Federal University Gusau, Gusau, Nigeria 15 12 2022 2022 2 12 27918 2 2022 18 11 2022 © The Author(s), under exclusive licence to Springer Nature Switzerland AG 2022, Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. Political speech acts are critical for politicians launching a regime because they can provide information that can be used to control people's thoughts and opinions. The purpose of this study was to conduct a qualitative content analysis of the inaugural and ascension addresses of Nigerian heads of state and presidents. The textual data used in this analysis were the ascension and inaugural addresses of Nigerian Heads of State and Presidents from 1960 to 2019. They were extracted and analysed using text-mining techniques. Textual data were clustered about their topical content using Latent Dirichlet Allocation (LDA), and speech cohesion between these addresses was examined using a similarity matrix and heatmap. Furthermore, term frequency and association analyses were performed to examine the high-frequency terms (tokens) and the terms (tokens) that are strongly correlated within each of the ascension/inaugural addresses (corpus). The summarization of characters and words in the ascension and inaugural addresses reveals that the Civilian Presidents used more characters and words than the Military Heads of State. There was an increase in the number of characters and words in the ascension and inaugural addresses among those who had served the nation multiple times. The total sentiment score in the ascension/inaugural addresses from 1960 to 2019 by Civilian Presidents and Military Heads of State revealed that the Civilian Presidents expressed more trust, surprise, sadness, joy, fear, disgust and anticipation in their addresses than the Military Heads of State. The most occurring term (token) in the ascension/inaugural addresses was the word government which appeared 221 times. The most token in the corpus government was found to be moderately correlated with the following tokens: loss, existing and majority. Similarly, economic was found to be moderately correlated with these tokens: inflation, building, education, exchange, loan, workers and technical. In this study, all the ascension/inaugural addresses share similar topic distribution: as seen in Abacha’s and Muritala’s addresses; and Shonekan’s inaugural address was very similar to Balewa, Azikwe and Babangida's addresses; Babangida's ascension, Abdulsalam’s 1998 ascension, Jonathan’s 2010 inaugural and Buhari’s 2015 inaugural addresses discussed similar topics to Obasanjo’s 1976 ascension address. The highest average sentiment score was observed in Obasanjo’s 2003 inaugural address and the lowest score was in Buhari’s 1983 ascension address. The sentiment score for the ascension/inaugural addresses showed that Civilian Presidents inaugural addresses expressed more positive, joy, trust and anticipation than Military Heads of State. These emotions showed that the Civilian President’s inaugural addresses are better when compared to Military Heads of State in terms of the sentiment scores. Keywords Data mining Sentiment analysis Head of State Nigeria Content analysis issue-copyright-statement© Springer Nature Switzerland AG 2022 ==== Body pmcIntroduction Language has played a pivotal role in gaining maximum control of many people’s thoughts, actions, opinions and values. People that crave power, most especially, political office holders have seen proper usage of the language as a medium of expressing their manifestos to the citizens, starting from the campaign to gain their vote of confidence, till the inauguration. The inaugural address is an avenue for giving the citizens' policy outlines penned down in the master plan for the new administration. Nigeria has had fifteen different heads of state and presidents since gaining her independence in 1960. The country is currently in its fourth republic: the first was from 1963 to 1966, the second from 1979 to 1983, the third from 1992 to 1993, and the fourth, and longest, from 1999 to the present. Every republic marks the beginning of the democratic dispensation era. It is, however, customary for every government (civilian or military) as well as the President or Head of State to deliver a political speech. The political speech ceremony (inauguration) into newly elected executive positions like president and governor, usually, in Nigeria, takes place on the 29th May of the election year since the fourth republic (return to democratic government after military disruptions), except for a few states like Kogi, Bayelsa, Ondo, Anambra and Edo due to the pronouncement by the court. Political speech is paramount for politicians initiating a regime as it provides information to regulate people's thoughts and opinions through political language. Permana and Mauriyat (2021a, b) and Charteris-Black (2018) classified political speech into policy making such as political decisions and putting in place shared values like consensus building. In the work of Osisanwo (2017) found in Ellah (2022), an inaugural speech is a speech delivered by every individual occupying a new political office during their ceremonial induction. Ellah and Nta (2020) stated in their study that political occupants not only express their qualities and traits during the inaugural speeches, but also declare their intentions and make promises to citizens. Several works have been studied in inaugural speeches in Nigeria. Adeyanju (2006) scrutinized the pragmatic features in the political speeches of six prominent Nigerian leaders: civilian leaders (Alhaji Abubakar Tafawa Balewa, Chief Obafemi Awolowo and Dr. Nnamdi Azikiwe); and military leaders (General Yakubu Gowon, General Olusegun Obasanjo and General Ibrahim Babangida). Akinwotu (2013) examined the political party nomination acceptance speeches of two Nigerian presidential candidates: Chief Moshood Kashimawo Olawale (MKO) Abiola of the Social Democratic Party (SDP) in the 1993 general election and Chief Obafemi Awolowo of the Unity Party of Nigeria (UPN) in the1979 election. Ellah (2022) observed that President Mohammadu Buhari’s 2015 inaugural speech from a discourse-pragmatic perspective, with a special interest in the incorporation and inculcation of other texts in the speech. Adegbija (1995) studied the discourse tactics in military coup speeches in Nigeria while seeking public approval. A comparative pragmatic investigation of the second inaugural speeches of Nigeria’s President Olusegun Obasanjo and United States’ President George Bush was also investigated, the work reveals that those having the same use of language and persuasive words in their speeches (Adetunji 2009). Sentiment analysis has been used in wide areas of study like e-commerce, communication, use of opinion mining to predict election outcomes and most recently coronavirus pandemic. Pak and Paroubek (2010) employed TreeTagger for the collected corpus of Twitter data to train a sentiment classifier based on the multinomial Naïve Bayes classifier. A sentiment classifier can identify positive, negative and neutral sentiments in documents. Yang et al. (2020) proposed a novel sentiment analysis model named Sentiment Lexicon on Chinese Based and Deep Learning (SLCABG) based on the sentiment lexicon and deep learning techniques to improve the sentiment analysis on product reviews. It was concluded that the new model can be used to assist merchants on e-commerce platforms to get feedback from users. Sentiment analysis has lately been used during the coronavirus pandemic. Sentiment analysis was applied to see how Indians felt towards the imposition of lockdown. It was revealed that most Indians' responses seem positive and follow the imposition of lockdown by the government to flatten the curve (Barkur et al. 2020). Onyenwe et al. (2020) used Natural Language Processing (NLP) technique referred to as sentiment analysis and tweets text exploratory to empirically examine the impact of political party control on winning an election over its candidates and vice versa. This was like the study of Oyebode and Orji (2019) that used lexicon-based and supervised machine learning approaches to election-related posts on social media to identify negative or positive sentiment polarity. According to Onyenwe et al. (2020), it was concluded that factors that play significant roles in winning an election were the influences of the political party and the candidate's disposition. The act of dissecting texts of speeches (especially political figures such as an executive head of state) has recently become a must-have for every household. There are two main techniques used, these are the assignment of predefined categories to texts (that is, text classification model); and the act of identifying specific information from the text (i.e. text extractor). Studies that used sentiment analysis in inaugural speeches can be found in Aremu (2017); Balogun and Amodu (2018); Enyi (2016b); Nnamdi-Eruchalu (2017); Ogungbe (2021); Osisanwo 2017). Ogungbe (2021) examined the lexico-syntactic expressiveness in President Muhammadu Buhari's 1983 and 2015 inaugural speeches and found that President Muhammadu Buhari employed stylistic devices like reference, harmony of words (collocation), enumeration, pronouns etc. to win the attention, support, trust and loyalty of Nigerians to the ideas expressed. The pragmatic and goals-oriented acts were identified in the use of words in President Muhammadu Buhari's 2015 inaugural speech in the investigation of Osisanwo 2017) using Mey (2001) pragmatic acts theory for descriptive analysis. Balogun and Amodu (2018) analysed the patterned repetitions that draw the attention of the audience in the inaugural speeches of President Goodluck Jonathan in 2011 and President Barack Obama in 2013. Being that President Obama is from an English-speaking nation and his speech was prepared by a skillful orator made his speech has an edge over President Jonathan in terms of being captivating and compelling. A word embedding (a subset of supervised machine learning technique) that can detect similar words and levels of negativity in Austrian parliamentary speeches from 1996 to 2013 was used. It was found that there is a greater usage of negative words in the opposition parties than in ruling parties (Rudkowsky et al. 2018). Text mining and sentiment analysis have been used in various facets of life but are limitedly explored in the inaugural speeches, especially in Nigeria. In the work of Han and Lim (2021), inaugural speeches delivered by 59 US Presidents from the two major political parties were classified using named entities and key phrases by using the Support Vector Machine (SVM). Han and Lim (2021) and Lim and Han (2020) used various N-grams lexical features rather than using topics or semantics to build a feature set. In this work, the ascension and inaugural addresses of Nigerian Heads of State and Presidents from 1960 to 2019 were analysed using text-mining techniques because only a few (if at all) works have been carried out using text mining and sentiment analysis on Nigeria’s presidents’ inaugural speeches. Governance is a very big challenge in the world. Nigeria is currently faced with a large distrust between the citizens and the leaders (Botha and Abdile 2019). The importance of this research work is to know the sentiment scores of Nigerian leaders and compare Civilian Presidents and Heads-of-State inaugural speeches using sentiment analysis. The methods in this study can also be applied to any country in the world with a military transition to democracy or vice versa. Methods The textual data used in this analysis are the ascension and inaugural addresses of Nigerian Heads of State and Presidents from 1960 to 2019, which were accessible online at www.dawodu.com. It is the first website about Nigeria's socio-economic problems, politics and history. Dr. Segun Dawodu started it as a blog for the first time in 1996. The data were saved into a CSV file format given that preprocessing of data is an integral part of any analysis task. The inherent structure of the ascension and inaugural addresses was analysed using text-mining techniques, which provide an efficient and reliable way of quantifying textual data. Textual data were clustered about their topical content using Latent Dirichlet Allocation (LDA), and speech cohesion between these addresses was examined using a similarity matrix and heatmap. R statistical software version 4.1.0 (2021-05-18)—"Camp Pontanezen" was used for the analysis. Results Distribution of characters & words in the ascension/inaugural addresses We performed some simple calculations and tidied the textual data by counting the number of characters (Speech Length) and words (Speech Words) in each of the ascension and inaugural addresses, with the longest address having 28,267 characters and 4,725 words. The summarization of characters and words in the ascension and inaugural addresses reveals that the Civilian Presidents used more characters and words than the Military Heads of State (Table 1).Table 1 Distribution of characters & words in the speech of the ascension/inaugural addresses Head N Sum of characters Sum of words Mean characters used Mean words used Civilian 10 146,649 24,137 14,665 2,414 Military 8 92,104 14,562 11,513 1,820 Specifically, the highest and the lowest characters for civilian presidents were found in Azikwe’s 1963 address which consists of 28,267 characters and in Shonekan’s 1993 address, which consists of 3909, respectively (Table 2). Also, there is an increase in the number of characters and words in the ascension and inaugural addresses among those who have served the nation multiple times. Obasanjo has the highest mean number of characters and words (Table 3).Table 2 Distribution of sum of characters & words in the ascension/inaugural addresses by name (alphabetically) and year Name Year Time spent in the office Sum of characters Sum of words Abacha 1993 4yrs, 203 days 16860 2685 Abdulsalam 1998 355 days 24396 3768 Azikiwe 1963 2yrs, 107 days 28267 4725 Babangida 1985 7yrs, 364 days 11823 1883 Balewa 1960 8yrs, 138 days 6863 1147 Buhari 1983 1 yr, 239 days 8501 1330 Buhari 2015 4yrs 11617 1915 Buhari 2019 4yrs* 19014 3021 Gowan 1966 8yrs, 362 days 13353 2110 Ironsi 1966 194 days 4915 772 Jonathan 2010 5yrs, 24 days 13202 2213 Muritala 1975 199 days 7062 1135 Obasanjo 1976 3yrs, 230 days 5194 879 Obasanjo 1999 4yrs 17466 2837 Obasanjo 2003 4yrs 25098 4160 Shagari 1979 4yrs, 91 days 12888 2056 Shonekan 1993 83 days 3909 669 Yaradua 2007 2yrs, 341 days 8325 1394 The * indicates the incumbent president Table 3 Distribution of mean characters & words in the ascension/inaugural addresses Name Number of time served Mean characters used Mean words used Buhari 3 13,044 2089 Obasanjo 3 15,919 2625 Corpus preparation and document-term matrix creation To prepare the corpus used for this analysis, we converted all text in the corpus to lower case, removed punctuation, numbers, stop words and stemmed words. The result is that each ascension and inaugural address is a string of tokens, where a token is a sequence of characters that are grouped as a useful semantic unit that is sometimes not always immediately recognizable as words. A document-term matrix was then created where each column represents a token and each row represents ascension and inaugural addresses. We created a document-term matrix, keeping only tokens longer than three characters since shorter tokens are very hard to interpret. The document’s term matrix has 89% sparsity (zero cells) with 4,921 terms (tokens) (Figs. 1, 2).Fig. 1 Distribution of the number of characters in the ascension/inaugural addresses Fig. 2 Distribution of the number of Words in the Ascension/Inaugural Addresses by Head Similarly, terms that appear in the collection of ascension and inaugural addresses less often than the specified lower bound were ignored, and terms that appear only once in the whole corpus were also ignored when building the second document-term matrix. This resulted in 1,889 terms (tokens) with a sparsity of 73%. The number of times each unique term (tokens) appears within each cell in the first document-term matrix was used to create a word cloud visualization to see the words occurring most frequently (Fig. 3), and the second-word cloud was created with the second document-term matrix (Fig. 4). Figure 4 becomes clearer than Fig. 3 after removing terms appearing in the collection of ascension and inaugural addresses less often than the specified lower bound.Fig. 3 Word cloud from the first document-term matrix Fig. 4 Word cloud from the second document-term matrix Sentiment analysis of the ascension/inaugural addresses The computational study of the sentiments and emotions expressed in the ascension and inaugural addresses was examined. There is a package in the R programming language called “syuzhet”. It is designed to classify the sentiments into positive, negative, fear, joy etc., by extracting the sentiments from the text. These sentiment words are available in the sentiment lexicon. The sentiment extraction method was developed by Standford for Natural Processing Language (Jockers 2020). The process searches each address for the appearance of certain words that are scored individually, which produces values that are marked as either exhibiting a positive or negative sentiment. The total Sentiment Score in the ascension/inaugural addresses from 1960 to 2019 by Civilian & Military Heads of State/Presidents shown in Fig. 5 reveals that the Civilian Presidents express more Trust, Surprise, Sadness, Joy, Fear, Disgust and Anticipation in their addresses than the Military Heads of State (Fig. 5). Obasanjo’s 2003 inaugural address has the highest positive and trust emotions, followed by Azikwe’s inaugural address in 1963. Surprisingly, Abacha’s ascension address in 1993 has no disgust emotion, while Obasanjo’s address in 1999 had the highest disgust emotion when compared with other Heads-of-State/Presidents' addresses (Fig. 6).Fig. 5 Total sentiment score in the ascension/inaugural addresses from 1960 to 2019 by civilian & military heads of state/presidents Fig. 6 Total Sentiment score in the ascension/inaugural addresses from 1960 to 2019 by name of heads of state/presidents and year The highest average sentiment score was observed in Obasanjo’s 2003 inaugural address, and the lowest score was in Buhari’s 1983 ascension address (Fig. 7).Fig. 7 Average sentiment Score by year of address and name of heads of state/presidents Exploring the document-term matrix Term frequency and association analyses were performed to examine the high-frequency terms (tokens) and the terms (tokens) that are strongly correlated within each of the ascension/inaugural addresses (corpus). The most occurring term (token) in the ascension/inaugural addresses was the government, which occurred 221 times. Other terms are presented in Table 4. The strength of the association (correlation) of the tokens with the twelve (12) most frequently occurring tokens in the ascension/inaugural address is presented in Table 5. Most token in the corpus government was found to be moderately correlated with the following tokens: loss, existing and majority. Similarly, economic was found to be strongly correlated with these tokens: inflation, building, education, exchange, loan, workers and technical ( See Table 5).Table 4 High-frequency tokens in the ascension/inaugural addresses (corpus) Frequency Terms (Tokens) 221 Government 150 Country, government, Nigeria 100 Country, government, Nigeria, Nigerians 80 Country, government, nation, national, Nigeria, Nigerians, people, political 60 Country, government, nation, national, Nigeria, Nigerians, people, administration, federal, military, political, economic 50 Country, government, nation, national, Nigeria, Nigerians, people, public, world, administration, continue, federal, fellow, military, political, economic Table 5 Token associations in the ascension/inaugural addresses (corpus) occurring 60 times (limited to the 10 tokens with the strongest associations with correlations >  = 0.65) Token Associated Tokens (Strength of Correlation) Country peaceful (.85), stability (.75), ongoing (.74), survival (.72), fellow (.71), culminating (.69), economic (.69), permanent (.69), chairman (.69), future (.68) Government loss (.69), existing (.68), majority (.67) Nation election (.81), foundation (.80), institutions (.77), industry (.76), individuals (.74), democracy (.71), firmly (.71), opportunity (.70), partners (.68), structure (.68) National institutions (.86), individuals (.81), peaceful (.80), true (.80), require (.79), provisional (.79), reconciliation (.79), ruling (.79), concern (.77), industry (.77) Nigeria assumption (.84), character (.84), coexistence (.84), criteria (.84), electorate (.84), ethics (.84), exercising (.84), fabric (.84), fourth (.84), hurt (.84) Nigerians healthcare (.86), inputs (.85), quantities (.85), subregion (.85), water (.85), output (.84), investment (.82), oil (.80), manufacturing (.80), fellow (.79) People understanding (.85), resolve (.78), required (.76), charge (.74), land (.74), administrations (.70), apply (.70), exploration (.70), real (.69), ensure (.68) Administration microeconomic (.99), funds (.98), abacha (.96), absence (.96), acceptability (.96), accommodation (.96), accomplish (.96), acknowledges (.96), acutely (.96), adjudication (.96) Federal corporation (.78), promulgated (.78), subject (.78), country (.71), council (.70), illegal (.67) Military corporation (.72), promulgated (.72), subject (.72), genuine (.69), region (.69), carry (.68), illegal (.68), provisions (.68), seize (.68), normal (.67) Political love (.87), structure (.81), constitutional (.80), committee (.80), institutions (.78), true (.78), time (.77), concern (.77), association (.76), judiciary (.76) Economic inflation (.81), building (.76), education (.68), exchange (.67), loan (.66), workers (.66), technical (.65) Topic modelling of tokens in the ascension/inaugural addresses (corpus) We perform topic modelling which is mixed-method modelling to uncover hidden thematic structures in the ascension/inaugural addresses (corpus). Before we dive into generating the topics and analysing the output, we ran a function to loop over different topic numbers to decide on the number of topics to use in the Latent Dirichlet Allocation (LDA) model. The model with the highest log-likelihood value between 2 and 50 topics indicates the optimum number of the topic that is the best fit for the corpus, in this loop 25 topics were chosen (Fig. 8). Table 6 shows the top-ranked 10 tokens associated with each of the 25 topics generated by the LDA. The topic association plot is a visualization showing the distribution of the 25 topics in each of the ascension/inaugural addresses (corpus) (Fig. 9).Fig. 8 LDA model selection results show the log likelihood of the corpus for different numbers of topics Table 6 Top-ranked 10 tokens associated with each of the 25 topics generated by the LDA model Topic 1 Topic 2 Topic 3 Topic 4 Topic 5 Topic 6 Topic 7 Topic 8 Topic 9 Topic 10 Topic 11 Topic 12 Topic 13 Government President Nigeria Nigerians Government Government Free Government Administration Law Political Government Government Joint Country Independence Country Federal People Determined Military Government Government National Economic Military People Citizens Day Nigeria Military Nigerians Nation Federation Political People Head Country Ensure Governance Build Nation Government Nigeria Boko Support Nigeria Nigerians Plans Chief Nigerians Democracy Federal Set Queen People Resources Haram Food Federal National Progress Dissolved International People Electoral Nigerian Representatives Fellow Nation Country Economic Region Parties Fellow Nigeria Nation Measures Country Nation World Challenges Nigerians Nigeria Policy Army Public Rebuilding Country Nigeria Past Chief Justice Constitutional Rural Political Nigerian Government Nigerian Elections Poverty Nation Military Public Power Fellow Country Economic Financial Forces World Regional Country Past Forces Nigerian Corruption Nigerians Property Friends Education Constitution National Housing Continue Nation Set Commanderinchief History National Topic 14 Topic 15 Topic 16 Topic 17 Topic 18 Topic 19 Topic 20 Topic 21 Topic 22 Topic 23 Topic 24 Topic 25 Government Country Nigeria Democracy Nigeria Nigeria Government Government Chief Government Africa Public Country Government Country Social Government Nigerians Military Country Col Country Commonwealth Nigeria Parties National Continue Sational Africa Country Coup Region Staff Education African Nigerians Federal Interim Nation Nigeria Power Leadership Public Military Government National Political Corruption Courage Economic Nigerian Process Freedom Nation Yakubu Eastern Nation Nigeria World Government Minister Nigeria People Africa Office Elections British Federal Federal Republic Office Administration Police Past Fight Nigerian World Democratic Country Comprising Forces Task Human Service Nigeria Sector Fellow Policies Person Note Nigerian Nigeria Armed Housing United People Dissolved Nation Women Continue National Vision Nigerians Nigerian Council God Colonial Country Chance Step Build Progress Nations Political Nigeria Revenue Military Nigerian Status Ensure Fig. 9 Topic association plot in the ascension/inaugural corpus Exploring cohesion in the ascension/inaugural addresses (corpus) using cosine similarity A cosine similarity analysis was performed to examine how each of the ascension/inaugural addresses is related to each other. In this current analysis, two addresses are similar if they share a similar topic distribution with a large cosine similarity measure between them. Although all the ascension/inaugural addresses are somewhat similar due to the large cosine similarity measure, Abacha’s address is very similar to Muritala’s address (0.91005), Shonekan’s inaugural address was very similar to Balewa’s (0.82407), Azikwe’s (0.85611), and Babangida’s (0.85140) addresses. Also, Babangida's ascension address (0.81303), Abdulsalam’s 1998 ascension address (0.83005), Jonathan’s 2010 inaugural address (0.83656) and Buhari’s 2015 inaugural address (0.81213) are very similar to Obasanjo’s 1976 ascension address (Table 7).Table 7 Cosine similarity matrix of the ascension/inaugural addresses (corpus) S/N Names Balewa Azikwe Gowon Ironsi Muritala Obasanjo Shagari Buhari Babangida Abacha Shonekan Abdulsalam Obasanjo Obasanjo Yaradua Jonathan Buhari Buhari 1 Balewa 1 2 Azikwe 0.77086 1 3 Gowon 0.7209 0.65882 1 4 Ironsi 0.72109 0.79367 0.80652 1 5 Muritala 0.81701 0.76941 0.78317 0.72597 1 6 Obasanjo 0.72796 0.71113 0.78317 0.78697 0.73357 1 7 Shagari 0.75403 0.79584 0.74317 0.73068 0.86751 0.71403 1 8 Buhari 0.75873 0.76163 0.74081 0.80036 0.78968 0.75439 0.71258 1 9 Babangida 0.77575 0.77629 0.78896 0.76706 0.69774 0.81303 0.78932 0.67964 1 10 Abacha 0.77339 0.74462 0.81466 0.76923 0.91005 0.71294 0.82751 0.80416 0.699 1 11 Shonekan 0.82407 0.85611 0.69158 0.75548 0.73195 0.73937 0.78081 0.7486 0.8514 0.73937 1 12 Abdulsalam 0.77701 0.66136 0.7524 0.70353 0.74118 0.83005 0.71348 0.75819 0.77882 0.68181 0.70896 1 13 Obasanjo 0.77068 0.66027 0.7571 0.65484 0.78552 0.7933 0.7886 0.72163 0.70733 0.78534 0.71095 0.85032 1 14 Obasanjo 0.71113 0.82624 0.79529 0.7314 0.78027 0.75113 0.87511 0.68833 0.8333 0.8114 0.82733 0.73647 0.75403 1 15 Yaradua 0.68253 0.81937 0.75493 0.82842 0.78262 0.73357 0.74715 0.86371 0.62462 0.79638 0.69701 0.69828 0.71493 0.74027 1 16 Jonathan 0.75457 0.75222 0.73557 0.68995 0.7714 0.83656 0.78534 0.71873 0.82914 0.76633 0.80217 0.86697 0.79113 0.79348 0.67656 1 17 Buhari 0.7638 0.75493 0.77629 0.85919 0.73104 0.81213 0.76851 0.8067 0.71747 0.76344 0.84706 0.7095 0.80941 0.74281 0.79548 0.70643 1 18 Buhari 0.74081 0.84833 0.71529 0.78244 0.70407 0.72977 0.80977 0.63819 0.79131 0.7229 0.80416 0.7457 0.74715 0.8543 0.73774 0.82281 0.74136 1 Clustering in the ascension/inaugural addresses (corpus) using a heatmap We used the cosine similarity matrix constructed earlier to map the different similarities in the ascension/inaugural addresses on a heatmap and visualize groups of addresses that are more likely to cluster together using the default hierarchical clustering method for the heatmap function. In our current analysis, a yellow square indicates strong similarity in addresses and a red square indicates dissimilar addresses. The dendrogram in Fig. 10 shows the steps in the hierarchical clustering algorithm and is similar and confirms the results of the cosine matrix output in Table 7. The numbers in Fig. 10 represent the presidents(see Table 7).Fig. 10 Heatmap of similarities between the ascension/inaugural addresses Discussion In this research, text-mining techniques were used to analyse textual data and sentiment. We have investigated the publicly accessible ascension and inaugural addresses of Nigerian Heads of State and Presidents from 1960 to 2019. These techniques stipulate a significant tool to appraise the consequences and expectations of Nigerians from a political speech. The sentiments of the speech were answered based on the results of the ascension and inaugural addresses of Nigeria's Heads of State and Presidents. The sum of characters for the civilian and military were 146,649 and 92,104 respectively. The sum of words used by the civilian president and military heads of state were 14,665 and 11,513 respectively. The average number of words used by the civilian presidents and military heads of state was 1820. We found out that the civilian presidents used more words and characters than the military heads of state. Shonekan, in the year 1993, used the lowest number of characters of 3909 and the number of words of 669 in the ascension/Inaugural addresses. Azikiwe in 1963 used the highest number of characters of 28,267 and several words of 4725. This is expected because of his work experience in the journalism industry. In 2003, Obasanjo used 25,098 characters with 4160 words, this was unexpected because of his military background, though he served as a civilian president in 2003. In 1983, Buhari used 8501 characters and 1330 words, in 2015, he used 11,617 characters and 1915 words, and in 2019, he used 19,014 characters and 3021 words. By comparing the leaders who served the country three times, Buhari and Obasanjo fulfilled this condition, Obasanjo used more characters and words than President Buhari in the ascension/inaugural Addresses. The total sentiment score for the Civilian Presidents and Military Heads of States were compared. For the Civilian Presidents, the sentiment analysis for the ascension/inaugural addresses was based on the count and the graph. Most of the emotions from the speeches were positive, and the positive emotions were higher when compared to emotions that showed anticipation, disgust, fear, joy, sadness, surprise, trust and negative. For the Military Heads of States, the sentiment analysis for the ascension/inaugural addresses based on the count and the graph, positive emotions were not up to half, the positive emotions were really small. We still observed a few emotions that indicated anticipation, disgust, fear, joy, sadness, surprise, trust and negative emotions for the military regime. Generally, Civilian Presidents had more positive emotions than Military Heads of States. The civilian presidents expressed more joy and trust emotions. The sentiment score in the ascension/inaugural addresses from 1960 to 2019 showed that the Civilian Presidents express more positive, trust, joy and anticipation emotions in their addresses than the Military Heads of State. This is similar to Permana and Mauriyat (2021b), where the speaker intended to create a future. Obasanjo’s 2003 inaugural address recorded the highest positive and trust emotions followed by Azikwe’s inaugural address in 1963. Abacha’s ascension address in 1993 had no disgust emotion, while Obasanjo’s address in 1999 had the highest disgust when compared with other Heads-of-State/Presidents' addresses. The highest average sentiment score was observed in Obasanjo’s 2003 inaugural address and the lowest score was in Buhari’s 1983 ascension address. Our model findings showed that the civilian presidents used more words and characters in the ascension and inaugural addresses than the military heads of states. This is in support of Enyi (2016a), where President Buhari's utterances were direct when he was a military Head of State. The leaders who had served the nation had an increment in the number of words and characters in their ascension and inaugural addresses. The highest average words and characters were recorded by Obasanjo. The word cloud indicated the importance and the words that appeared frequently in the ascension and inaugural addresses of Nigerian Heads of State and Presidents. The word “government” was more frequent than other words in the ascension and inaugural addresses. Conclusion In this study, we were able to establish that the length and number of words in ascension/inaugural speeches delivered by civilian presidents are greater than the military heads of state. In addition, among past Presidents and Heads of State who had served at least twice, Obasanjo had the highest average number of characters (length of words) and number of words. The research also showed that the ascension/inaugural addresses share similar topic distribution: as seen in Abacha’s and Muritala’s addresses; and Shonekan’s inaugural address was very similar to Balewa, Azikwe and Babangida's addresses; Babangida's ascension, Abdulsalam’s 1998 ascension, Jonathan’s 2010 inaugural and Buhari’s 2015 inaugural addresses discussed similar topics to Obasanjo’s 1976 ascension. However, Obasanjo’s 2003 inaugural address has the highest positive and trust emotions followed by Azikwe’s inaugural address in 1963. Surprisingly, Abacha’s ascension address in 1993 has no disgust emotion while Obasanjo’s address in 1999 had the highest disgust emotion when compared with other Heads-of-State/Presidents' addresses. Furthermore, civilian presidents express more emotions such as Trust, Surprise, Sadness, Joy, Fear, Disgust and Anticipation in their addresses than the military heads of state in terms of the sentiment scores. The highest average sentiment score was observed in Obasanjo’s 2003 inaugural address and the lowest score was in Buhari’s 1983 ascension address. Acknowledgements The authors are grateful to the anonymous reviewers for their comments to improve the clarity and quality of the paper. Author contributions KRF contributed to the conceptualization, LOM contributed to the review of literature, KRF, LOM, ME, OJI contributed to the manuscript preparation. ME contributed to the logic and planning. KRF, LOM, ME, OJI contributed to the study design. KFR contributed to the data analysis. OJI contributed to the discussion of the findings. KRF, LOM, ME, OJI read and approved the final manuscript. Funding No funds, grants, or other support was received. Data availability The datasets analysed during the current study are available from www.dawodu.com. Declarations Conflict of interest On behalf of all the authors, the corresponding author states that there is no conflict of interest. Ethical approval We used secondary data which are available for public use. Informed consent This article does not contain any studies with human participants performed by any of the authors. ==== Refs References Adegbija E I, major-general X: discourse tacts in military coup speeches in Nigeria Interdiscip J Study Discourse 1995 15 2 253 270 Adetunji A Odebunmi A Arua AE Arimi S The Speech Acts and Rhetorics in the Second Inaugural Addresses of Nigeria’s President Olusegun Obasanjo and America’s President George W Bush Language Gender and Politics A Festschrift for Yisa Kehinde Yusuf 2009 Lagos Concept Publications Limited 275 296 Adeyanju D Pragmatic features of political speeches in english by some prominent nigerian leaders Ife Stud Eng Lang 2006 6 1 48 63 Akinwotu SA A speech act analysis of the acceptance of nomination speeches of chief obafemi awolowo and chief M.K.O. Abiola Eng Linguist Res 2013 2 1 43 51 Aremu, M. (2017) Pragmatic analysis of conceptual mappings in inaugural speeches of Nigerian presidents. Covenant J Lang Stud (CJLS) 5(2):32–47 Balogun S Amodu J Parallelism in Inaugural Speeches of Presidents Barrack Obama and Goodluck Jonathan 2018 Nigeria LAPAI RESEARCH IN HUMANITIES Barkur G Vibha, & Kamath, G. B. Sentiment analysis of nationwide lockdown due to COVID 19 outbreak: evidence from India Asian J Psychiatry 2020 51 102089 10.1016/j.ajp.2020.102089 Botha A Abdile M Reality versus perception: toward understanding Boko Haram in Nigeria Stud Confl Terror 2019 42 5 493 519 10.1080/1057610X.2018.1403152 Charteris-Black J Analysing political speeches: Rhetoric, discourse and metaphor 2018 Basingstoke Palgrave Macmillan Ellah SM Intertextuality in political discourse: a study of President Mohammadu Buhari’s 2015 inaugural speech J Lang Linguist Lit Stud 2022 11 1 9 17 Ellah SM Nta NG Illocutionary function of language in the healing miracles of jesus christ LWATI: J Contemp Res 2020 17 3 148 163 Enyi AU Pragmatic analysis OF Nigeria’s president Muhammadu Buhari’s maiden coup address of january 1, 1984 and his inaugural address of may 29, 2015: a comparative appraISAL Int J Eng Lang Linguist Res 2016 4 5 47 64 Enyi AU Pragmatic analysis of Nigeria’s president Muhammadu Buhari’s maiden coup address of January 1, 1984 and his inaugural address of May 29, 2015: a comparative appraisal Int J Engl Lang Linguist Res 2016 4 5 47 64 Han H Lim S Inaugural Speech classification with named entities and key phrases IEEE Int Conf Big Data Smart Comput (BigComp) 2021 10.1109/BigComp51126.2021.00029 Jockers, M. M. (2020). Package “syuzhet” Type Package Title Extracts Sentiment and Sentiment-Derived Plot Arcs from Text. https://github.com/mjockers/syuzhet Lim S Han H Distinguishing political parties from inaugural speeches using machine learning IEEE Conf Multimed Inf Proc Retr (MIPR) 2020 10.1109/MIPR49039.2020.00026 Mey JL Pragmatics: an introduction 2001 2 Oxford Blackwell Nnamdi-Eruchalu GI A critical discourse analysis of Muhammadu Buhari’s inaugural speeches with a focus on pronouns Int J Lit Lang Linguist 2017 4 2 156 164 Ogungbe EO (2021) Lexico-syntactic Stratagems in President Muhammadu Buhari’s Inaugural Speeches. ANSU J Lang Lit Stud (AJLLS) 2(1):1–16 Onyenwe, I., Nwagbo, S., Mbeledogu, N., Onyedinma, E., Onyenwe, I., Nwagbo, S., Mbeledogu, N., & Onyedinma, E. (2020). The Impact of Political Party/Candidate on the Election Results From A Sentiment Analysis Perspective Using #AnambraDecides2017 Tweets. https://courses.lumenlearning.com/americangovernment/chapter/introduction-9/ Osisanwo A I Belong to Everybody yet to nobody: pragmatic acts in president Muhammadu Buhari’s inaugural speech Athens J Mass Med Commun 2017 3 4 297 320 10.30958/ajmmc/3.4.2 Oyebode O Orji R Social Media and Sentiment Analysis: the Nigeria Presidential Election 2019 2019 Piscataway IEEE Pak A Paroubek P Twitter as a corpus for sentiment analysis and opinion mining Lrec 2010 10 1320 1326 Permana TJ Mauriyat A Analysis of speech acts in susilo bambang yudhoyono’s political speech Prof J Eng Educ 2021 4 1 62 68 Permana TJ Mauriyat A Analysis of speech acts in susilo bambang yudhoyono’s political speech PROJECT (prof J Eng Educ) 2021 4 1 62 Rudkowsky E Haselmayer M Wastian M Jenny M Emrich Š Sedlmair M More than bags of words: sentiment analysis with word embeddings Commun Methods Meas 2018 12 3 140 157 10.1080/19312458.2018.1455817 Yang L Li Y Wang J Sherratt RS Sentiment analysis for e-commerce product reviews in chinese based on sentiment lexicon and deep learning IEEE Access 2020 8 23522 23530 10.1109/ACCESS.2020.2969854

0

PMC9753883

NO-CC CODE

2022-12-16 23:26:20

no

SN Soc Sci. 2022 Dec 15; 2(12):279

utf-8

SN Soc Sci

10.1007/s43545-022-00570-x

oa_other

==== Front NZ J Educ Stud New Zealand Journal of Educational Studies 0028-8276 2199-4714 Springer Nature Singapore Singapore 272 10.1007/s40841-022-00272-1 Commentary Commentary: ‘Plumbing Leaks: The post-pandemic Neoliberal turn of Tertiary Students’ http://orcid.org/0000-0003-1206-9219 Olsen-Reeder Vini vini.olsen-reeder@vuw.ac.nz grid.267827.e 0000 0001 2292 3111 Te Kawa a Māui, Victoria University of Wellington, Wellington, New Zealand 15 12 2022 19 26 10 2022 15 11 2022 16 11 2022 © The Author(s) under exclusive license to New Zealand Association for Research in Education 2022, Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. ==== Body pmcIntroduction As the educator of an Indigenous language, I feel a particular draw to the ethics around inclusivity and accessibility. I must introduce my students to their heritage language in new ways that help them forge a path toward a stronger identity. In order to strengthen their identities, I need to consider the myriad of identity constructions that enter my classroom. This is why inclusivity and accessibility are core parts of my work. As a student-focussed educator, I therefore find it a strange position to sit in opposition to the general student voice. In recent discussions on campus and online around post-pandemic digital teaching avenues, I have been in this opposition, and in approaching a commentary about this topic, I ask the question: have tertiary students in New Zealand taken a rather remarkable neoliberal turn? In October 2022, the students’ association of my institution ran a campaign demanding that lecture recordings be universally guaranteed for all classes on campus (Victoria University of Wellington Students’ Association, 2022a). The campaign garnered a few thousand signatures which is a good result, but ultimately it was not successful in its aims. While I applaud the idea of challenging through petitions, I immediately felt uncomfortable about the assertions the petition seemed to make. The petition extolled the virtue of lecture recordings, in that “Universal access to lecture recordings is a cornerstone of accessible education” (VUWSA, 2022a). Inclusivity and accessibility seem to be a key catalyst for this petition, especially for students who have additional learning needs, and the recording of all classes on campus are pitched as the solve. The issue, of course, is that this assertion is not entirely true, and in my view, the result of entrenching those recordings could have been more harmful than the benefits they provide. For an educator who works closely with students from non-mainstream backgrounds, I have fought many battles to protect their learning needs. I have become somewhat of an expert in digital teaching post-pandemic and know that the road to inclusive and accessible education is far more complex and exciting than mere recordings. I further understand that digital teaching is paved with inequity and inaccessibility. Inspiring learning through technology while deconstructing its drawbacks is therefore crucial. While some of our students certainly benefit from lecture recordings, others are excluded. While some classes can provide recordings with little to no effort, other classes require hours to manage recordings safely (such as removing private conversation). Recordings are useful for many students, but for others, risk disengagement away from a course entirely. Given those things are all true, what then, is to be gained from universal lecture recordings? In this post-pandemic world where digital teaching and neoliberalism are intertwined, it makes sense to me to consider what other benefits could befall signatories of a petition like this. It would, for example, allow students to do other things such as work, go overseas on holiday, take on external commitments, or even stay in bed. I know these things do happen. In considering neoliberalism and ethics then, key questions are whether any of this is inherently based in inclusivity, and secondly, whether any the effects of those activities are fair on the person charged with carrying out that labour? I feel there is a risk that ideals around inclusive and accessible education have been co-opted to pursue other objectives, and that those things are possibly more based in neoliberal privilege and consumerism. Ultimately, I am drawing the conclusion that students have - perhaps unknowingly - taken a neoliberal turn in post-pandemic life. With students redefining themselves as consumers of trade in a retail knowledge context, there is room for the neoliberal co-option of that inclusivity and accessibility. This is a known part of neoliberalism, where “[w]hat is promoted ethically [i]s often at odds with what it need[s] concretely” (Bloom, 2017, p. 26). Universal recordings are pitched as the answer to inclusive and accessible education, but lecture recordings are far from that answer. Students – customers in a Retail Store? “I appreciate the equity, diversity and inclusion objectives that they are trying to achieve, but I think sometimes these can have unintended consequences” (participant, in Potter et al., 2022). Since 2020, my institution has required all classes to have lecture recordings, with a few exemptions. The future of those lecture recordings, post-pandemic, is not guaranteed, however. During 2022, the author’s institution has been discussing what digital pedagogical requirements will exist from around 2024 onwards, and the petition from the students’ association was the response.1 One underlying issue is how the nature of a petition matches up to the neoliberal context within which the petition sits. To return to Loyola-Hernández et al. (2022), the neoliberal “model turns students into consumers of a product delivered by staff.” They stress that:[w]e must not see students as consumers but as equal partners in this learning process. Yet, this is increasingly difficult to do as those of us working in HEIs will only be able to implement a “bounded” social change which is a “social change that is imaginatively bound by the constraints of the students’ immediate environment: the neoliberal university. (Loyola-Hernández et al., 2022, drawing from Connelly & Joseph-Salisbury, 2019, p. 1037) It is therefore possible to situate the petition as a neoliberal transaction between a consumer (the student) and a retailer (the university). If this is correct, and students now assert themselves as customers, we are now needing to explicate their presence on campus in a more neoliberal sense. This is scary - the collaboration we should be working through together, as educationalists and caretakers of human knowledge, is no longer important – only the terms of trade. As educators are neither the customer nor the manufacturer, there is room for the two parties to negotiate the tools used to provide the ‘good,’ but the teacher is not part of that conversation. A second complication is that neoliberalism and capitalism often contradict in ethic and reality. While the students’ association was battling for universal lecture recordings, it was concurrently running a F*CK THE BARE MINIMUM2 campaign, which asked questions such as, “are the lectures worth the course fees?” (Victoria University of Wellington Students’ Association, 2022b). I do sympathise - students are probably not getting the same ‘deal’ on education that they were prior to the pandemic. As Loyola-Hernández et al. (2022) state, complaints have arisen within: larger teaching scenarios such as seminars and lectures mainly due to students not being assessed for their classes or “not getting what they paid for”. A number of staff reported students felt cheated from their university experience and some had even “asked for a refund”. Students pay a large amount in fees under neoliberal education regimes. In a neoliberal sense, it is possible that, once treated like a customer, some students may indeed want to be that customer. In any other neoliberal context, if the product retailed is not what the customer expected, there is scope to demand a refund. What is difficult about this situation, is that the low-quality educational product currently being purchased by students, and the one the campaign is complaining about, is a direct result of our emergency pandemic response: lecture recordings. Yet, these are what the petition seeks to entrench permanently. The unintended consequences of eternalising recordings are twofold: firstly, they solidify the poor-quality students are questioning fee payments over, and secondly, they risk any real chance to convince an institution it should improve its inclusive teaching practises with real investment later on. Lecture recordings do diminish the quality of teaching in the classroom (Olsen-Reeder, 2022) and until we collaboratively pursue quality, inclusive digital teaching, the overall standard of the learning experience likely will not improve. Co-option and Performativity The freedom that neoliberalism offers […] turns out to mean freedom for the pike, not for the minnows. (Monbiot, 2016). In pandemic-related online teaching moves, it is true that “teachers reveal concerns with this modality of teaching and assessment, mentioning that it increases the existing social inequalities if the conditions for success are not ensured for all students” (Peixoto et al., 2022). While technology is certainly useful and could help to solve our inclusion and accessibility needs, it is not automatically the equity-access ‘silver bullet’ many make it out to be – that takes work outside of a pandemic. In Li and Yu’s (2022) literature review of other scholarly work, they find it an “insufficient substitute for classroom teaching.” Watermeyer and others (2022) review a number of studies which purport “various alleged inadequacies claimed of sectoral (and national) digital infrastructures – not to mention concerns related to the ‘unsatisfactory realities’ and ‘digital downsides’ of technology use in higher education.” The case is no different here in an Aotearoa context (Olsen-Reeder, 2022), where synchronous online and in-person, and asynchronous teaching during the pandemic have removed most markers of teaching quality, because they are both being offered simultaneously. The arguments for offering things like lecture recordings are mostly based around equality and inclusion, things that allow students with learning needs more freedom to navigate their education. The author supports these concepts entirely. What isn’t true though, is that universality of lecture recordings equates to equality or inclusion. Neoliberalism requires those on campus to devalue “empathy”, “compassion”, and “pro-social qualities … that hold people directly impacted… in the highest regard” when making decisions (Killam, 2022). Lecture recordings are fraught with inconsistencies and quality issues, as well as a myriad of follow-on issues that mean it can very quickly be a mistake (Olsen-Reeder, 2022). Digital learning can, therefore, be both low in quality, and high in inequality, and it is the devaluing Killam speaks of I now address. Issues span the gamut of real-life sociodemographic problems, to learning needs, to staff welfare issues. A blanket, catch-all universal mentality in any digital learning concept is likely to create more concern for marginalised people than it alleviates, and this is what leads my concern that we are making assumptions about inclusive learning that are not true. Udeogalanya (2022) notes that in the pandemic, students from lower socio-economic backgrounds are forced to find devices elsewhere to study digitally, or make risky purchases to keep up, that many cannot use these devices alone, and that all leads to poor performance and early exits. Loyola-Hernández and others (2022) purport that in pandemic-related teaching “the burden fell to individuals or grassroots initiatives rather than authorities, including HEIs” to keep students afloat, and that “For many of us it was impossible to continue to study and work at our universities while our fellow migrant community was struggling.” Similar can be said for students here in Aotearoa, including Māori students (New Zealand Union of Student’s Associations, 2020; Te Mana Ākonga, 2020). In students with learning needs, Kocdar & Bozkurt (2022) note that digital education presents a wave of new barriers to learners with additional needs if not executed well enough, and Arsantas & Gul (2022) note that there are particular ICT solutions students with visual impairment need to learn effectively. Classroom surveillance also causes anxiety for some students, as their contributions can be filmed and rewatched by their peers (Wang & Zhang, 2021). As a language educator who works with an Indigenous community of students, inclusivity and equality are at the fore of my energy expenditure each day. I can assert that language anxiety is present in my classroom for several reasons, one is that all elements of speech are being captured and broadcast for the class to see as a whole, on repeat. There are staff welfare issues to consider as well, as they are the ones who have carried this extra labour for three pandemic-ridden years, in the name of empathy and compassion for learners. They are, in the main, academic (teachers and tutors) and ICT staff, but there are administration staff involved too. It is broadly acknowledged that academic mothers have been particularly affected by the COVID-19 pandemic (França et al., 2022), and that “the ideal worker lying behind the neoliberal governance model – someone unencumbered by non-work demands – is grounded on gendered divisions” (Rosa, 2022). Burton and Bowman (2022) comment on the “carelessness and violence as part of the ‘academic precariat’ in that such oppressive actions and atmospheres continually push certain scholars (women, people of colour, disabled and working-class people) to the margins of academia, both professionally and intellectually.” That same ‘precariat’ is recalled by Collins et al. (2022) where it is noted that “tutors were aware of potential impacts of perceived negative student feedback on their precarious contractual situation.” There are the unsung heroes of online pandemic responses - ICT staff, “an historically marginalised constituency within universities, whose contribution is habitually far from understood, frequently confused and thus inadequately valued” (Watermeyer et al., 2022). All of these people - some of whom are student peers – are surely worthy of labour reconstruction away from tasks that contribute to their inequality. To order them to continue their emergency pandemic teaching response, for eternity, seems to me, the pinnacle of neoliberal thought, in that they have not been considered at all. A final point, I further note that for Māori, “Indigenous inclusion” within university contexts can be more performative than meaningful and real (Hoskins & Jones, 2022). I am often suspicious of the motives and agendas behind things claimed to be ‘inclusive’ of Māori, but are not Māori-designed, and are not welcoming of reasonable Māori thought. My community have been ignored in this petition, and so, I cannot see how it can be inclusive of the teaching practise my classroom needs, because my class has been excluded from the shaping of it as a whole. Is the Neoliberal Student Prepared for Itself? As we enter a new stage of digital learning, a key question is whether students themselves recognise how different their role must be as online learners, because the role of the student and educator changes significantly. Li and Yu (2022) note that online learning is exceedingly difficult for the student: Sustainable education challenged students psychologically via online technology in a developing society. Students were unlikely to ask questions due to cultural norms, forming a gap between them and teachers. Udeogalanya (2022) might agree, noting that in the pandemic their students “recounted the stress, fear, anxiety, depression, and insomnia that they were dealing with as they studied. Many complained that they had difficulty focusing and concentrating on their coursework. They worried that they would fail their examinations.” Of course, the pandemic will have contributed to these negative experiences, but educators know there is much more to consider in long-haul digital learning, and if students are not ready, these experiences will remain long after the pandemic. Studying from home is otherwise named online “student-centered learning (SCL)”, and it “encourages students to take more responsibility for their learning” (Núñez-Canal, et al., 2022, citing O’Neill & McMahon, 2015), where they are both “autonomous and responsible” (Núñez-Canal, et al., 2022, citing McCabe and O’Connor, 2014). Online SCL is not just about watching a video and completing assessments as if the student was present in class – the entire ethos of learning changes. This includes assessment, where, as Devlin and Samarawickrema (2022) establish in their literature review, students “direct their own learning” and “assess the quality of their own work” carried out by “using contemporary digital tools.” In this alignment, “[t]he educator conveys knowledge and is the architect of an educational process, ranging from generating content, designing a learning experience, or accompanying a student during a discovery” (Núñez-Canal et al., 2022). It is possible that students have not yet been party to the full gamut of information needed to make sound pedagogical digital change, in-and-of themselves. Udeogalanya (2022) points out that “[f]aculty are encouraged to focus on how to hone students’ technology skills by increasing student engagement, assessment, and feedbacks, thus creating effective and meaningful learning,” but it seems unfair that a student should enter a course without prior warning that the learning context has changed from anything they have ever known before. It is necessary too, to have a mechanism to redirect students who are unwilling to contribute in this way to another learning context, possibly at another institution. It is not clear whether this is understood. The lack of clarity is seen in instances where it is hard to continue “nurturing and maintaining staff-student relationships in a remote context”, and Schalk et al. (2022) note that “[c]ontinuing these social relationships with students was a challenge, and it was noted that students were hesitant to contribute verbally in synchronous classes, preferring to use chat feature.” A small example of passive learning behaviour online, but a destructive mechanism towards collaborative engagement, none-the-less. Ultimately, anything overly ‘passive’ in online HE SCL, like a chat box, is unlikely to succeed. Wekerle (2022) points out that because “all these activities typically entail a rather passive role of students,” this propels “doubts on higher education teachers’ abilities to use the potentials of digital technology to promote high-quality student learning.” Since being a passive SCL learner is the antithesis of good learning behaviour, students would need to expect more from themselves as we institute HE SCL more permanently – and educators would need to expect this from them as well – if the whole idea was going to work. As well as doubling the workload for educators (remember, there are still fee-paying students in the room who need to be taught), online SCL requires an incredibly high level of digital fluency on the part of the student, and it would be a mistake to assert that even students who are digital natives had this capacity innately. To draw from a participant in Schalk et al. (2022): I think we think that the students are very digitally literate, and they’re not … they’re very digitally literate in terms of some aspects of social media … we sort of have this expectation that they are cottoned on to it and know exactly what to do and that’s not the case. In another study, Vishnu et al. (2022) note that “more than one-third of the students were found to be moderately competent with regard to their ability to adapt and mix various media for study purposes, protect personal data and privacy, protect health and wellbeing while using digital technologies, and understand the impact of digital technologies on environment.” That leaves two thirds who did not have any of these skills necessary to partake safely and equitably in online learning. To summarise, learning off-campus requires students to enter an HE SCL course with a completely different mindset, and with the understanding they will have to work hard to acquire the digital learning skill necessary to engage in a course. Then, they will need to coerce themselves to participate actively. This is, of course, in addition to other aspects covered in other sections of this paper, such as doubling course workloads, and the myriad of equity issues online SCL raises for certain groups of students. Even if educators managed all of these things, would online SCL students do their parts, too? Conclusion Nordbäck calls on a quote by Ashcraft (2017, in Nordbäck, 2022): “you bustle among the tenants and habitats of a Neoliberal U whose projects include an increasingly neoliberal you.” Managerialist and marketized policies and practices make the “neoliberal us” work more and they force us to manage (for) ourselves.” When I’ve raised these issues, my concerns have generally been met with feelings my worries are ‘not the student’s job to sort out.’ This, though, seems to fit the idea that an individual’s neoliberal standing shall “be blamed on those considered to be unworthy individuals who have not provided for themselves in times of need” (Benatar et al., 2018). It aligns with the description offered to us by Ashcraft, that an individual must ‘manage-for’ themselves in the face of neoliberal change, and it further fits the construction of the “care-less academy” as painted by Burton and Bowman (2022). My disagreeing too, is perhaps also seen as hurtful to a customer ‘who is always right.’ I have to be frank though, disagreeing is an important part of being a critic and conscience of society. I’m no stranger to making a point to protect my community, and if I believe I possess the prerequisite knowledge to make an informed point, I will make it. To look past these things seems, to me, like the customer telling the plumber how to plumb. Eventually, the tap will leak. Declarations Conflict of interest There are no conflicts of interest to note. 1 To see the petition, visit: https://www.change.org/p/guarantee-students-universal-access-to-lecture-recordings. 2 To see this campaign, visit: https://www.vuwsa.org.nz/fck-the-bare-minimum. Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References Arslantas T Gul A Digital literacy skills of university students with visual impairment: a mixed-methods analysis Education and Information Technologies 2022 27 5605 5625 10.1007/s10639-021-10860-1 35068987 Benatar S Upshur R Gill S Understanding the relationship between ethics, neoliberalism and power as a step towards improving the health of people and our planet The Anthropocene Review 2018 5 2 155 176 10.1177/2053019618760934 Bloom, P. (2017). Producing the “Ethical” Capitalist Subject. In The Ethics of Neoliberalism, (1st ed., pp. 19–36). Routledge. 10.4324/9781315619019-2 Burton S Bowman B The academic precariat: understanding life and labour in the neoliberal academy British Journal of Sociology of Education 2022 43 4 497 512 10.1080/01425692.2022.2076387 Collins H Glover H Myers F Behind the digital curtain: a study of academic identities, liminalities and labour market adaptations for the ‘Uber-isation’ of HE Teaching in Higher Education 2022 27 2 201 216 10.1080/13562517.2019.1706163 França, T., Godinho, F., Padilla, B., Vicente, M., Amâncio, L., & Fernandes, A. (2022). Having a family is the New Normal”: parenting in Neoliberal Academia during the COVID-19 pandemic. Gender Work & Organization, 1–17. 10.1111/gwao.12895. Hoskins T Jones A Indigenous inclusion and indigenising the University New Zealand Journal of Education Studies 2022 10.1007/s40841-022-00264-1 Killam, R. (2022). My high horse is dying: Agitating Internalized Neoliberalism in Higher Education with(Out) Compassion. Cultural Studies ↔ Critical Methodologies, 0(0), 10.1177/15327086221107050. Kocdar, S., & Bozkurt, A. (2022). Supporting learners with special needs in Open, Distance, and digital education. In O. Zawacki-Richter, I. Jung (eds), Handbook of Open, Distance and Digital Education, pp.1–16. DOI: 10.1007/978-981-19-0351-9_49-1 Li M Yu Z Teachers’ satisfaction, role, and Digital literacy during the COVID-19 pandemic Sustainability 2022 14 3 1121 10.3390/su14031121 Monbiot, G. (2016). Neoliberalism – the ideology at the root of all our problems. Retrieved 26 October 2022, from: https://www.theguardian.com/books/2016/apr/15/neoliberalism-ideology-problem-george-monbiot New Zealand Union of Students’ Associations (2020). COVID-19 and Tertiary Students: The impact on the wellbeing, finance, and study of students at tertiary institutions in Aotearoa New Zealand. Retrieved 11 October 2022, from: https://static1.squarespace.com/static/5f0515b1b1a21014b5d22dd6/t/5fa218b297ec03254a8fc9b3/1604458682311/COVID-19+and+Tertiary+Students+The+impact+on+the+wellbeing%2 C+finance%2 C+and+study+of+students+at+tertiary+institutions+in+Aotearoa+New+Zealand+.pdf Nordbäck E Hakonen M Tienari J Academic identities and sense of place: a collaborative autoethnography in the neoliberal university Management Learning 2022 53 2 331 349 10.1177/13505076211006543 Olsen-Reeder, V. (2022). Dual-mode teaching in the language classroom: reconciling the pandemic, equity, and the future of quality language teaching pedagogy. New Zealand Journal of Educational Studies, 57(2), Doi: 10.1007/s40841-022-00258-z. Peixoto, P., Almeida, J., & Albuquerque, C. (2022). When Assessment Moves Home: The Digital Panopticon in Higher Education. In: Mesquita, A., Abreu, A., Carvalho, J. (eds), Perspectives and Trends in Education and Technology. Smart Innovation, Systems and Technologies, vol 256. Singapore: Springer. 10.1007/978-981-16-5063-5_43 Rosa, R. (2022). The trouble with ‘work–life balance’ in neoliberal academia: a systematic and critical review. Journal of Gender Studies, 31(1), DOI: 10.1080/09589236.2021.1933926. Te Mana Ākonga (2022). Impacts of the COVID-19 Lockdown on Māori learners in Private Training Establishments Retrieved 11 October 2022, from: https://www.temanaakonga.org.nz/nga-puka Udeogalanya, V. (2022). Aligning digital literacy and student academic success: Lessons learned from COVID-19 pandemic. International Journal of Higher Education Management, 8(2), DOI: 10.24052/IJHEM/V08N02/ART-4. Victoria University of Wellington Students’ Association (2022a). Petition to Guarantee Students Universal Access to Lecture Recordings. Retrieved 25 October, 2022, from: https://www.vuwsa.org.nz/media/2022/10/4/0mr44ammtorwl9i80kgknbs4e222ir F*CK THE BARE MINIMUM. Victoria University of Wellington Students’ Association, & Retrieved (2022b). 11 October, 2022, from: https://www.vuwsa.org.nz/fck-the-bare-minimum Wang, X., & Zhang, W. (2021). Psychological anxiety of College Students’ Foreign Language Learning in Online Course. Frontiers in Psychology, 12(598992), DOI: 10.3389/fpsyg.2021.598992. Watermeyer R Crick T Knight C Digital disruption in the time of COVID-19: learning technologists’ accounts of institutional barriers to online learning, teaching and assessment in UK universities International Journal for Academic Development 2022 27 2 148 162 10.1080/1360144X.2021.1990064

0

PMC9753884

NO-CC CODE

2022-12-16 23:26:20

no

NZ J Educ Stud. 2022 Dec 15;:1-9

utf-8

oa_other

==== Front Braz. J. Chem. Eng. Brazilian Journal of Chemical Engineering 0104-6632 1678-4383 Springer International Publishing Cham 291 10.1007/s43153-022-00291-x Original Paper Methanogenic consortia from thermophilic molasses-fed structured-bed reactors: microbial characterization and responses to varying food-to-microorganism ratios Fuess Lucas Tadeu ltfuess@alumni.usp.br 12 Eng Felipe felipe.eng@usp.br 2 Bovio-Winkler Patricia patricia.bovio@gmail.com 3 Etchebehere Claudia cetchebehere@iibce.edu.uy 3 Zaiat Marcelo zaiat@sc.usp.br 2 Nascimento Claudio Augusto Oller do oller@usp.br 1 1 grid.11899.38 0000 0004 1937 0722 Chemical Engineering Department, Polytechnic School, University of São Paulo. Av. Prof. Lineu Prestes, 580, Bloco 18—Conjunto das Químicas, São Paulo, SP 05508-000 Brazil 2 grid.11899.38 0000 0004 1937 0722 Biological Processes Laboratory, São Carlos School of Engineering, University of São Paulo (EESC/USP), Av. João Dagnone 1100, São Carlos, SP 13563-120 Brazil 3 Microbial Ecology Laboratory, Department of Biochemistry and Microbial Genomics, Biological Research Institute “Clemente Estable”, 3318 Italia Avenue, Montevideo, Uruguay 15 12 2022 121 24 7 2022 20 11 2022 5 12 2022 © The Author(s) under exclusive licence to Associação Brasileira de Engenharia Química 2022, Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. The heterogeneous character of fixed-film reactors may create highly specialized zones with a stratified distribution of microbial groups and varying capabilities to withstand high organic loads in anaerobic digestion (AD) systems. The microbial distribution and methane-producing potential of biomass from different regions (feeding zone and structured bed) of two second-stage thermophilic (55 ºC) fixed-film reactors were assessed. Three levels of food-to-microorganism (F/M) ratio (0.4, 1.0 and 3.0 g-COD g−1VS) using fermented (two-stage AD) and fresh (single-stage AD) sugarcane molasses were tested in batch reactors, simulating low to high organic loads. Specific methane production rates increased as the F/M increased when using fermented molasses, maintaining efficient methanogenesis at substrate availability levels threefold higher than single-stage schemes (3.0 vs. 1.0 g-COD g−1VS). Success in methane production derived from the homogenous establishment (similar in both feeding zone and bed) of syntrophic associations between acetogens (Pelotomaculum, Syntrophothermus, Syntrophomonas and Thermodesulfovibrio), acetate oxidizers (Thermoacetogenium, Mesotoga and Pseudothermotoga) and hydrogenotrophic methogens (Methanothermobacter and Methanoculleus) replacing acetoclastic methanogens (Methanosaeta). Phase separation under thermophilic conditions was demonstrated to boost methane production from sugar-rich substrates, because the process depends on microbial groups (hydrogenotrophs) that grow faster and are less susceptible to low pH values compared to acetotrophs. Supplementary Information The online version contains supplementary material available at 10.1007/s43153-022-00291-x. Keywords Two-stage biodigestion Sugarcane molasses 16S rRNA gene amplicon sequencing Kinetic assessment Syntrophic acetate oxidation + hydrogenotrophic methanogenesis http://dx.doi.org/10.13039/501100001807 Fundação de Amparo à Pesquisa do Estado de São Paulo 2017/00080-5 2015/50684-9 2015/06246-7 2014/50279-4 Fuess Lucas Tadeu Zaiat Marcelo Nascimento Claudio Augusto Oller do http://dx.doi.org/10.13039/501100002322 Coordenação de Aperfeiçoamento de Pessoal de Nível Superior PROEX Zaiat Marcelo ==== Body pmcIntroduction Anaerobic digestion (AD) of high-strength wastewaters has been facilitated by developing high-rate systems, in which biomass retention mechanisms enabled uncoupling the hydraulic retention time (HRT) from the solid retention time in reactors (van Lier et al. 2015). Biomass granulation or the provision of stationary or moving media for cell attachment has created suitable conditions for the application of high organic loading rate (OLR) under low HRT, resulting in a series of benefits, such as achieving higher robustness and operating stability, requiring more compact units and increasing biogas production. Among the numerous configurations proposed are two that successfully achieved effective full-scale applications, namely, the sludge blanket and fixed-film systems (Moletta 2005). The vulnerability of sludge granulation to compositional characteristics commonly found in numerous wastewaters, such as high suspended solid content and high salinity (van Lier et al. 2015), provides some advantages to using support materials in cell retention. Focus is given to the second group in this study, because recent research efforts have led to improvements on the conventional packed-bed systems through vertically arranging the support media to characterize the anaerobic structured-bed reactor (AnSTBR) (Aquino et al. 2017; Camiloti et al. 2014; Fuess et al. 2021a; Mockaitis et al. 2014). AnSTBR systems simultaneously provide high void indices and adequate surface area levels for biomass attachment, which expand the applicability of fixed-film reactors. Compared to packed-bed systems, the higher bed porosity (> 90%) (Aquino et al. 2017; Mockaitis et al. 2014) minimizes bed clogging-related limitations, which enables both the processing of solid-rich substrates and the establishment of microbial populations with high growth yields (e.g. hydrolytic and fermentative bacteria). The attached growth itself also comprises an advantageous feature compared to other high-rate systems because less metabolic energy is required to establish high cell densities within the system (Aquino et al. 2017; Blanco et al. 2017). Consequently, the microbial community presents higher robustness when faced with adverse environmental conditions, such as organic overloading events (Chan et al. 2009). Interestingly, fixed-film reactors do not behave as homogeneous systems, because the low-to-null cell mobility associated with the progressive conversion of the substrate may create highly specialized zones in the reactor. Following the flow direction, a “more acidogenic” microbial community should be expected in regions with high fresh substrate availability (e.g. feeding zones), whilst a “more acetogenic” and mainly methanogenic-like character should be observed in the subsequent compartments of the reactor. It is worth highlighting that the fermentability of the substrate may directly impact this microbial distribution, because the proposed specialization should be more evident in the processing of easily fermentable substrates, such as sugar-rich ones. Conversely, a relatively more homogeneous microbial distribution (or at least without the prevalence of fermentative groups near the feeding zone) is expected to occur in second-stage methanogenic reactors, in AD schemes where acidogenesis is carried out separately from methanogenesis. Different microbial populations will have different growth rates (Mosey 1983), potentially leading to wide variations of the food-to-microorganism (F/M) ratio along fixed-film reactors. The amount of microorganisms is commonly measured as volatile suspended solids (VSS) or only volatile solids (VS), whilst the “food” is represented by the organic matter content measured as the chemical oxygen demand (COD). Recent studies associated optimal fermentation activity (targeting biohydrogen production) with F/M ratios within the range of 4.0–6.0 g-COD g−1VSS d−1 (Blanco et al. 2017; Anzola-Rojas et al. 2015; Fuess et al. 2021b; Hafez et al. 2010), whilst stable methanogenesis has been conventionally pointed to occur under much lower F/M ratio values (< 0.4 g-COD g−1VSS d−1 (Aquino et al. 2017; Barros et al. 2016; Chernicharo 2007), specifically in single-stage schemes. However, the compartmentalized assessment of a second-stage AnSTBR indicated the establishment of highly efficient substrate conversion into methane under F/M ratios as high as 3.0 g-COD g−1VSS d−1 (Fuess et al., 2021a) directly contradicting the expected behavior for methanogenic environments. In practical aspects, high F/M ratios indicate (or simulate, in the case of batch reactors) situations of organic overloading (or close to that), in which the activity of fermentative bacteria should prevail and inhibit methanogens by providing an excess amount of volatile fatty acids (VFA). Synonyms for the term F/M ratio in continuous reactors include both “biological loading rate” and “sludge loading rate” (Chernicharo 2007), as well as “specific organic loading rate” (Anzola-Rojas et al. 2015). Some important questions arise from the stratified character of fixed-film reactors, regarding both the spatial distribution of specialized microbial groups and variations in the quality of the substrate (driven by the pre-fermentation intensity): [i] Does biomass specialization effectively occur as a result of differences in substrate quality (following the flow direction) or is a homogeneous distribution of all microbial groups participating in substrate conversion repeated along the entire reactor? [ii] More specifically, can methanogens withstand higher organic loads (or higher F/M ratios) when phase separation is used, i.e. when substrate fermentation is carried out (separately) prior to methanogenesis? To answer these questions, the microbial distribution and methane-producing potential of biomass collected in different regions of two second-stage thermophilic AnSTBR systems were assessed. Samples collected from the feeding zone and the bed region were compared using 16S rRNA gene sequencing to understand performance discrepancies as a function of the prevailing microbial groups. Batch tests using biomass from the feeding zones were further conducted under different F/M ratio values using fermented and fresh (non-fermented) sugarcane molasses as the substrate to understand how efficiently methanogenic consortia are able to perform under low and excess substrate availability levels, potentially defining operating limits for anaerobic reactors. Material and methods AnSTBR systems and thermophilic methanogenic consortia Two bench-scale (1.9 L each) AnSTBR systems were sampled to obtain the methanogenic consortia microbiologically characterized and used as inocula in methane production potential (MPP) tests in this study. Polyurethane (PU) foam strips wrapped in spiral polyvinyl chloride (PVC) frames were used as the support material in both reactors, which were operated in upflow mode and fed with fermented sugarcane molasses for 250 (RM1) and 230 (RM2) days under thermophilic temperature conditions (55 ºC). The reactors were subjected to different alkalinization strategies: while sodium bicarbonate (NaHCO3) dosing was carried out during the entire operating period of RM1, sodium hydroxide (NaOH) dosing coupled to effluent recirculation was initially used in RM2 (185 d), after which this approach was replaced by NaHCO3 dosing. Increasing OLR levels were applied to both systems to define the operating limit conditions for each alkalinization strategy. Complete details of the long-term operation of RM1 and RM2 were previously presented elsewhere (Fuess et al. 2021a). Despite the differences aforementioned, the operation of both reactors was finalized under equivalent conditions of OLR (10.0 kg-CODt m−3 d−1; CODt = total chemical oxygen demand), HRT (24.0 h) and influent CODt (10 g L−1), in addition to NaHCO3 dosing. Biomass samples were further collected during reactor disassembling separately from the feeding zone (suspended cells in the bulk liquid) and from the bed region (PU strips—attached cells). Attached biomass was removed from the media by washing the strips with distilled water. Sample cleansing prior to DNA extraction was carried out immediately after collection, whilst the remaining amounts of biomass from the feeding zones were stored at 4 ºC prior to use in MPP tests. Figure 1 shows constructive aspects of the AnSTBR and details of the biomass collection during the disassembling procedure.Fig. 1 Details of the AnSTBR systems (simplified sketch and experimental prior to inoculation) and biomass sampling from the feeding zone (FDZ) and structured bed (STB) during reactor disassembling. The same procedure was applied in RM1 and RM2. Legend: 1- reactor feeding, 2- effluent collection, 3- biogas collection Sugarcane molasses characterization Fresh sugarcane molasses (Mol) was collected from a large scale annexed biorefinery (milling approximately 10 million tons of sugarcane per harvest) located in Pradópolis, São Paulo, Brazil. The samples were stored at 4 ºC prior to use in both the feeding of the acidogenic reactor (see further description) and MPP tests. Fermented molasses (fMol) was collected from a bench-scale (2.0 L) thermophilic (55 ºC) acidogenic AnSTBR operated under optimized conditions (OLR = 86.0 kg-CODt m−3 d−1; HRT = 10.0 h) targeting enhanced biohydrogen production, as reported elsewhere (Fuess et al. 2021b). fMol samples were stored at −20 ºC prior to use (feeding both RM1 and RM2 and as substrate in MPP tests) to interrupt the activity of cells washed-out from the acidogenic reactor and preserve the compositional aspects of the fermented substrate. Detailed compositional aspects of Mol and fMol are presented in Table 1.Table 1 Compositional characterization of fresh and fermented molasses Parameter Sugarcane molasses Fresh (Mol) Fermented (fMol) CODt (mg L−1) 707 ± 10a,b 30,900 ± 260 CODs (mg L−1) 707 ± 10a,b 28,203 ± 927 CHt (mg L−1) 546 ± 19a (81.9%) 1717 ± 148 (6.5%) Lactate (mg L−1) – 13,942 ± 601 (52.9%) Ethanol (mg L−1) – 329 ± 22 (2.4%) Acetate (mg L−1) – 1204 ± 160 (4.6%) Propionate (mg L−1) – 101 ± 4 (0.5%) Butyrate (mg L−1) – 2656 ± 168 (17.1%) Caproate (mg L−1) – 169 ± 31 (1.3%) PheOH (mg L−1) 13.0 ± 0.1a (2.1%) 667 ± 24 (2.7%) SO42− (mg L−1) 2.4 ± 0.1a 470 ± 5 CODs/SO42− (-) 299.6 60.0 ± 1.9 pH (-) – 5.18 ± 0.07 CODt total chemical oxygen demand, CODs soluble chemical oxygen demand, CHt total carbohydrates, PheOH total phenols, SO42−- sulfate aValues in mg g−1 bCODt = CODs. Percentage values (%) indicate the relative participation of the compound in the CODs Methane production potential tests MPP tests were carried out in 500 mL (nominal volume) Erlenmeyer flasks (with DIN standard PBT screw caps—GL45) filled with 100 mL of substrate and inoculum. Both Mol and fMol were diluted to obtain a COD of 10 g L−1, simulating the terminal operating condition applied in the continuous reactors. Sugars were promptly solubilized once Mol was diluted. Meanwhile, fMol was centrifuged (9000 rpm, 5 min) prior to dilution as a strategy to remove cells (fermentative biomass) remaining from the acidogenic reactors, i.e. only the supernatant was used. In practical aspects, only the soluble organic content of both Mol and fMol was used as the substrate; however, the nomenclature “soluble COD (CODs)” was not used in this specific step because no 0.45 μm-filtration was carried out. Conversely, methanogenic sludge sources were centrifuged under similar conditions (9000 rpm, 10 min) to remove excess liquid phase, i.e. only the pellets were used, which were called wet sludge. In addition to biomass samples collected from the feeding zones of RM1 and RM2, the thermophilic methanogenic sludge used in the inoculation of both continuous reactors (identified as IN) was also assessed in MPP tests, following the same pre-centrifugation step. Assessing only biomass samples from the feeding zone was based on the performance of the continuous reactors, because the CODs was majorly consumed (> 70%) in this compartment in both reactors (Fuess et al. 2021a). Once the volatile solids (VS) content was measured in each methanogenic consortium, the amount of wet sludge added to each flask was calculated according to three different F/M (or substrate-to-inoculum) ratios, namely, 0.4, 1.0 and 3.0 g-COD g−1VS. The selected F/M ratios were based on the specific organic loading rate (sOLR) assessed in the continuous reactors, specifically in RM1 (Fuess et al. 2021a): the values 0.4 and 1.0 g-COD g−1VS simulated the sOLR (obtained in terms of g-COD g−1VSS d−1) observed under stable operating conditions, using the total amount of biomass inside the reactor and the amount of biomass retained specifically in the feeding zone as the references, respectively. In turn, the value of 3.0 g-COD g−1VS represented the maximum sOLR observed in the feeding chamber still associated with a stable operation. The three levels of the F/M ratio were assessed in fMol-fed reactors using all sludge sources (RM1, RM2 and IN). In the particular case of Mol, only sludge from RM2 was used. Finally, blank reactors were also monitored (duplicates) by replacing Mol or fMol with distilled water, using a fixed mass of 10 g of wet sludge per flask. For identification purposes, the nomenclature used to differentiate the conditions was formed by three parts: F/M ratio value-inoculum source(substrate type)—e.g. 0.4-RM1(fMol) and 3.0-RM2(Mol). The adopted experimental arrangement enabled comparing two- and single-stage AD schemes. In addition to the definitions regarding both the F/M ratio and inoculum sources, reactor alkalinization (including blanks) was based on NaHCO3 dosing (0.25 g-NaHCO3 g−1COD), also using the continuous reactors as the reference (Fuess et al. 2021a). pH values in fMol and Mol after NaHCO3 dosing reached ca. 6.70 and 7.60, respectively. After filling the flasks with alkalinized substrate and sludge, the reactors were fluxed with ultra-high purity grade nitrogen for 5 min to establish conditions of anaerobiosis and sealed with rubber stoppers and plastic caps. The reactors were incubated in an orbital shaker (model Multitron PRO Incubator Shaker—Infors HT, Infors AG, Bottmingen-Basel, Switzerland) and monitored for periods ranging between 7 and 16 d, depending mainly on the temporal evolution of methane production. Temperature and agitation conditions were set respectively as 55 ºC and 100 rpm. Reactor monitoring was carried out through periodically sampling both the gas and liquid phases with 1 mL insulin syringes. In the specific case of biogas collection, the syringe was equipped with a Teflon body two-way valve (Supelco™ Analytical—Sigma Aldrich, Bellefonte, PA, USA). The sampling periodicity varied according to the response of the systems in terms of the cumulative methane production and COD decay. Experimental runs were finalized after observing coefficient of variation (CV) values lower than 5% for at least three consecutive points of the cumulative methane production, whilst the stabilization of COD decay was used as a secondary reference. Liquid phase sampling was limited to 10% of the total substrate volume (10 mL). Biogas production was monitored by measuring the internal pressure of the flasks followed by the compositional analysis of the sample. Pressure measurements were carried out using a pressure gauge model TPR-16 (Desin Instruments, s.a., Barcelona, Spain), with an upper measurement limit of 500 mbar (50 kPa). Whenever required, pressure reliefs were conducted to prevent the occurrence of pressure values out of the measurement range. Analytical methods, calculations and kinetic analysis Overall liquid phase measurements included determining the pH, CODt, CODs, total carbohydrates (CHt), lactate, total phenols (PheOH), partial (PA) and intermediate (IA) alkalinity, volatile organic acids by titration (VOAtit), volatile organic acids (VOAGC; C2–C6) and solvents by gas chromatography (GC) and sulfate (SO42−). The periodic monitoring of the liquid phase in MPP tests included exclusively CODs measurements, whilst all remaining parameters were measured in non-diluted substrates (Table 1; except for PA, IA and VOAtit) and by the end of the incubation periods (except for the CODt). COD (both total and soluble), pH and SO42− measurements were based on the Standard Methods for the Examination of Water and Wastewater (APHA et al. 2012), whilst specific protocols were used for the remaining parameters: CHt (Dubois et al. 1956), lactate (Taylor 1996), PheOH (Buchanan and Nicell 1997), PA/IA (Ripley et al. 1986), VOAtit (Kapp 1984) and VOAGC/solvents (Adorno et al. 2014). Except for the case of CODt determination, samples were filtered in 0.45 μm syringe membranes (Chromafil GF/PET, Macherey–Nagel GmbH & Co. KG, Düren, Germany) prior to all remaining analyses. Considering gas phase monitoring, biogas composition (nitrogen—N2, methane—CH4, carbon dioxide—CO2 and hydrogen sulfide—H2S) measurements were carried out in a GC equipped with a thermal conductivity detector (model GC-2014, Shimadzu Scientific Instruments, Japan) using hydrogen as the carrier gas (Lebrero et al. 2016). Specifically, the cumulative methane production up to time “n” (VCH4,n; NmL) was calculated as proposed elsewhere (Santos et al. 2019) using Eq. (1), in which the terms iPpre-inj,n, iPpost-inj,n-1, fCH4,n, Vheadspace, T0, P0 and Texp, are, respectively, the internal pressure prior to sampling at time “n” (atm), the internal pressure after sampling at time “n—1” (atm), the methane fraction in biogas at time “n” (dimensionless), the volume of the headspace (L), the temperature at normal conditions (273.15 K), the pressure at normal conditions (1 atm) and the incubation temperature (55 ºC = 328.15 K).1 VCH4,n=∑t=0niPpre-inj,n-iPpost-inj,n-1×fCH4,n×Vheadspace×T0P0×Texp The modified Gompertz model (Zwietering et al. 1990) (Eq. 2) was fitted to VCH4 values to obtain the potential methane production (PCH4; NmL), the maximum methane production rate (RCH4; NmL h−1) and the lag phase period (λ; h). The maximum methane production rate was also calculated in specific terms (RCH4’; NmL h−1 g−1VS) by dividing RCH4 by the amount of sludge (VS) used in the inoculation of each reactor. In Eq. (2), the terms VCH4(t) and e are, respectively, the cumulative methane production (model) as a function of the incubation period (t) and the Euler’s number (2.71828). Endogenous methane production (assessed in blank reactors) was discounted from methane production values obtained when processing fMol and Mol. Hence, VCH4 values (both experimental and fitted) correspond to the net methane production. In parallel, kinetics of substrate consumption (CODs) was assessed through fitting the first-order decay model with residual (Rodrigues et al. 2003) (Eq. 3) to CODs temporal profiles, from which the residual CODs (CODsR; mg L−1), the initial CODs (CODs0; mg L−1) and the first-order kinetic constant (k1; h−1) were estimated. The specific first-order kinetic constant (k1’; h−1 g−1VS) was calculated analogously to RCH4’. Model fitting was carried out using the software Origin 2020 (OriginLab Corporation, Northampton, MA, USA) using the Levenberg–Marquardt algorithm. In the particular case of methane production, models were adjusted to experimental points collected up to fulfilling two criteria: [i] cumulative CODs removal efficiency (ERCODs; Eq. 4) higher than 70% and [ii] CODs removal efficiency between two consecutive points (ERCODs,interval; Eq. 5) lower than 5%. In Eq. (4)–(5) the terms CODsn, CODs0 and CODsn-1 are the CODs (mg L−1) measured at times “n”, 0 (zero) and “n – 1”, respectively.2 VCH4t=PCH4·exp-expRCH4·ePCH4λ-t+1 3 CODst=CODsR+CODs0-CODsR×exp-k1×t 4 ERCODs,n=100×CODsn-CODs0CODs0 5 ERCODs,interval=100×CODsn-CODsn-1CODs0 Microbial community characterization by 16S rRNA gene sequencing Biomass sampling and sample preparation. Preliminary details of the microbial community sampling (targeting microbial characterization) were presented in Section “AnSTBR systems and thermophilic methanogenic consortia”. Samples from suspended cells in the feeding zone and attached cells from the bed region were obtained for both continuous methanogenic reactors (Fig. 1) during reactor disassembling and promptly stored at – 20 ºC after cleansing, as described in the sequence. The thermophilic sludge (IN) used in the inoculation of RM1 and RM2 (collected from a thermophilic full-scale UASB processing sugarcane vinasse) was also characterized. Because the COVID-19 pandemic forced a relatively prolonged storage (4 ºC) of the remaining biomass samples from the feeding zones (ca. 6 months) and from IN (ca. 14 months) prior to MPP tests, aliquots of the stored samples were also analyzed (except for the case of RM1). Hence, a total of seven samples were processed and were identified by the sludge source (IN, RM1 or RM2) and the reactor region, i.e. feeding zone (FDZ) or structured bed (STB). The nomenclature “ps” (post-storage) was also used when pertinent. Table 2 brings details of the timeline followed to obtain biomass samples. All samples were subjected to 2–3 rounds of centrifugation (6000 rpm, 10 min) using a phosphate-buffered saline, which aimed to remove impurities (residual organic compounds). Finally, biomass pallets were stored at – 20 ºC prior to further processing.Table 2 Timeline detailing biomass sampling periods. The text in bold refers to the nomenclature of the samples Experiment Timeline Event Description Continuous reactors (Fuess et al., 2021a) Before day 1 Inoculation Reactor filling with macerated sludge Operation in closed cycle for 4 d Cleansing and storage (-20 ºC) of one biomass sample from the inoculum (IN) for microbial characterization Storage (4 ºC) of the remaining aliquot of the UASB-derived sludge Until day 230 Continuous operation of RM2 No biomass sampling Day 230 Disassembly of RM2 Reactor drainage, cleansing and storage (-20 ºC) of biomass samples collected from the feeding zone (RM2-FDZ) and from the bed region (RM2-STB) of RM2 Storage (4 ºC) of the remaining aliquot of biomass collected from the feeding zone of RM2 Until day 250 Continuous operation of RM1 No biomass sampling Day 250 Disassembly of RM1 Reactor drainage, cleansing and storage (-20 ºC) of biomass samples collected from the feeding zone (RM1-FDZ) and from the bed region (RM1-STB) of RM1 Storage (4 ºC) of the remaining aliquot of biomass collected from the feeding zone of RM1 Activity interruption due to COVID-19 pandemic (~ 180 d) Batch tests (this study) – – Cleansing and storage (-20 ºC) of samples from the long-term stored inoculum (INps) and from the long-term stored biomass collected in the feeding zone of RM2 (RM2-FDZps) for microbial characterizationa aAn unsuccessful preliminary round of batch tests using biomass collected from the feeding zone in RM1 decreased the availability of this sludge source, so that no sample was taken after the long-term storage in this particular case DNA extraction, 16S rRNA gene amplicon sequencing and bioinformatics The procedures used in DNA extraction and 16S rRNA gene sequencing are presented elsewhere (Piffer et al.  2022)—refer to Supplementary material for details. Sequences were analyzed using quantitative insights into a microbial ecology’ pipeline (QIIME2 2021.4 release) (Bolyen et al. 2019). Demultiplexed paired-end sequencing reads were imported into the QIIME2, resulting in 358,837 joined sequences. Sequencing reads were ‘denoised’ using the ‘divisive amplicon denoising algorithm’ DADA2 (Callahan et al. 2016) plugin in QIIME2, which provided high resolution amplicon sequence variants (ASV) for downstream analysis by filtering out noises, correcting errors in marginal sequences, removing chimeric sequences and singletons and dereplicating the resulting sequences. As a result, 313,539 sequences were obtained, ranging from 39,937 to 46,692 per sample (n = 7 in total), representing 719 ASV. The consensus sequences for the ASV were classified with a classify-sklearn classifier trained against the most recent SILVA 16S rRNA gene reference (release 138.1) database (Quast et al. 2013). In order to complete downstream diversity and composition analyses, sequences were rarefied to the lowest number of sequences per sample (n = 39,937 sequences). The sequences were submitted to the National Center for Biotechnology Information (NCBI; http://ncbi.nlm.nih.gov) under accession BioProject ID PRJNA778433. Data analysis and visualization All the statistical analyses were performed in R version 3.5.1 with R Studio environment, Version 1.3.1093. The biom file from QIIME2 was imported and analyzed through phyloseq-modified workflow (McMurdie and Holmes 2013). Taxon relative abundance bar charts were generated using custom R scripts and ggplot2 (v3.3.2) (Wickham 2016). A similarity matrix of weighted UniFrac distances was used for ordination by principal coordinates analysis (PCoA) of non-transformed relative abundance diversity datasets implemented in phyloseq and plotted using ggplot2 to visualize the distribution of the microbial composition. Heat maps were generated using ampvis2 (v.2.6.5) (Andersen et al. 2018). Results and discussion Kinetics of methane production The methane production potential of thermophilic microbial consortia processing sugarcane molasses was assessed under different F/M ratios, simulating different levels of substrate availability and also comparing single- and two-stage AD schemes. Temporal profiles for the cumulative methane production and CODs decay (experimental data) are depicted in Figs. 2 and 3, respectively. The kinetic parameters of the fitted models describing methane production and CODs decay are detailed in Tables 3 and 4, respectively. RCH4 values followed an increasing pattern as the F/M ratio decreased (Table 3), regardless of the inoculum source and substrate type. This pattern characterizes an expected result, because higher biomass amounts (lower F/M ratios) will have the ability to promptly convert the available substrate. However, an inverse pattern was observed when considering RCH4’ values, which increased as the F/M ratio was increased for a given inoculum source (Table 3; considering specifically the use of fMol). This result indicates that lower amounts of biomass produced methane more efficiently, most likely due to the favoring of substrate conversion kinetics under excess substrate availability (prior to reaching the limit concentration, i.e. CODs, prior to inhibition). Under conditions of excess biomass (lower F/M ratios), a fraction of the cell most likely remained in a latency-like state, without effectively participating in methane evolution. From a kinetic perspective, lower substrate availability was the limiting factor for methanogens.Fig. 2 Temporal evolution of methane production (VCH4, experimental data) for F/M ratios of a 0.4 g-COD g−1VS, b 1.0 g-COD g−1VS and c 3.0 g-COD g−1VS Fig. 3 Temporal profiles of CODs decay (experimental data) for F/M ratios of a 0.4 g-COD g−1VS, b 1.0 g-COD g−1VS and c 3.0 g-COD g−1VS Table 3 Kinetic models fitted to experimental methane production profiles Substrate Sludge source F/M ratio Kinetic parameter R2 PCH4 RCH4 RCH4'a λ (NmL) (NmL h−1) (NmL h−1 g−1VS) (h) Fermented molasses RM1 0.4 207.11 ± 2.85 13.91 ± 0.55 5.68 10.84 ± 0.73 0.99 1.0 173.38 ± 10.91 8.11 ± 0.60 7.70 26.63 ± 1.80 0.99 3.0 165.48 ± 9.37 3.91 ± 0.11 11.91 46.92 ± 1.20 0.99 RM2 0.4 301.30 ± 7.76 9.34 ± 0.24 3.69 21.39 ± 0.89 0.99 1.0 279.24 ± 7.96 5.78 ± 0.14 5.70 40.62 ± 1.30 0.99 3.0 234.71 ± 5.37 3.76 ± 0.05 11.13 68.27 ± 0.84 0.99 IN 0.4 235.10 ± 14.86 4.94 ± 0.18 1.96 102.65 ± 1.90 0.99 1.0 207.48 ± 13.61 3.43 ± 0.13 3.39 120.65 ± 2.38 0.99 3.0 183.92 ± 5.58 2.16 ± 0.04 6.42 142.01 ± 1.67 0.99 Fresh molasses RM2 0.4 263.81 ± 20.36 7.69 ± 0.60 3.08 65.05 ± 2.99 0.99 1.0 298.75 ± 27.43 2.03 ± 0.08 2.00 102.50 ± 6.46 0.97 3.0b – – – – – PCH4 potential methane production, RCH4 maximum methane production rate, RCH4' specific maximum methane production rate, λ- lag phase period aCalculated using average values bNo model was fitted in this condition Table 4 Kinetic models fitted to experimental CODs decay profiles Substrate Sludge source F/M ratio Incubation period (h) Kinetic parameter R2 CODs0 CODsR k1 (× 10–2) k1' (× 10–2)a (mg L−1) (mg L−1) (h−1) (h−1 g−1VS) Fermented molasses RM1 0.4 0–163 9817 ± 136 1566 ± 81 3.392 ± 0.155 1.384 0.99 1.0 24-163b 8050 ± 528 1764 ± 414 3.058 ± 0.787 2.904 0.95 3.0 72-163c 6525 ± 211 2027 ± 239 3.767 ± 0.687 11.478 0.95 RM2 0.4 0–192 10,308 ± 182 996 ± 267 1.456 ± 0.121 0.575 0.99 1.0 0–48 10,188 ± 152 6525 ± 806 2.766 ± 1.102 2.728 0.96 48–192 7817 ± 316 349 ± 630 1.537 ± 0.319 1.516 0.96 3.0 0–96 10,046 ± 197 5867 ± 1751 1.109 ± 0.759 3.282 0.91 96–216 7209 ± 162 659 ± 289 2.014 ± 0.231 5.960 0.99 IN 0.4 106-336d 8566 ± 258 2066 ± 159 2.415 ± 0.254 0.956 0.97 1.0 106-336d 9061 ± 307 961 ± 382 1.357 ± 0.189 1.343 0.96 3.0 125-216d 8604 ± 78 5144 ± 166 2.489 ± 0.297 7.401 0.99 216–384 5351 ± 164 1091 ± 225 1.620 ± 0.248 4.817 0.97 Fresh molasses RM2 0.4 0–334 10,204 ± 504 888 ± 547 1.182 ± 0.221 0.473 0.96 1.0 0–216 9723 ± 245 3632 ± 223 2.112 ± 0.274 2.083 0.96 216–438 3737 ± 178 1031 ± 153 2.954 ± 0.997 2.995 0.96 3.0 0–334 9945 ± 330 4678 ± 197 2.307 ± 0.386 6.827 0.91 CODs0 initial CODs, CODsR residual CODs, k1 first-order kinetic constant, k1' specific first-order kinetic constant aCalculated using average values bLinear decay between 0 and 24 h cLinear decay between 0 and 72 h: CODs(t) = (9796 ± 107)—(43.999 ± 2.381) × t (R2 = 0.97) dThe CODs consumption was not significant until the lower limit of the interval was reached An integrated analysis including RCH4’ values obtained for RM1-derived sludge (this study) and specific methane production rate values in the different operating phases of RM1 (Fuess et al. 2021a) revealed a consistent variation pattern for RCH4’ (Fig. 4a–b), regardless of the operating mode, i.e. batch or continuous. The specific methane production rate in RM1 (at the different OLR levels) was calculated through re-arranging performance data previously presented, namely, the volumetric methane production rate and the biomass retention, providing a novel response not addressed in the base reference study (Fuess et al. 2021a). Details of this calculation are presented in the Supplementary data section. Despite the different measurement units for the F/M ratio (g-COD g−1VS) and the sOLR (g-COD g−1VSS d−1), both parameters measure the amount of substrate available for conversion in the reactors on an equivalent basis, which supports the integrated comparison. RCH4’ varied linearly up to the maximum experimental sOLR measured in RM1 (~ 5.0 g-COD g−1VSS d−1; Fig. 4a), indicating that F/M ratio values assessed in batch tests (≤ 3.0 g-COD g−1VS) characterized conditions of excess sludge compared to the amount of substrate, justifying the increasing trend previously highlighted. Differently from batch systems, in which the initial F/M is defined and fixed to a desired value, the sOLR varies as a function of both the applied OLR and the amount of retained biomass within continuous reactors (Fuess et al. 2021a; Anzola-Rojas et al. 2015), explaining the varying values in RM1. Interestingly, specific methane production was subjected to some degree of inhibition with the increase in the OLR (Fig. 4b), but not as a direct consequence of excess substrate availability. With the continuous retention of biomass within the continuous system, increasing the OLR (up to 20 kg-COD m−3 d−1) was not sufficient to proportionally increase the sOLR (Fig. 4b), explaining the lower RCH4’ values. Hence, the F/M ratios assessed in the batch tests may represent two operating situations of the continuous reactors: a short-term operation characterized by a low OLR under equally low biomass retention levels or a long-term operation with relatively high OLR and excess sludge retention.Fig. 4 Correlations between methane production and substrate availability: a maximum specific methane production rate (RCH4’) as function of the F/M ratio (batch reactors) and specific organic loading rate (sOLR, continuous reactor) and b RCH4’ as function of the organic loading rate (OLR) and sOLR (continuous reactor). Note: aValues calculated through re-arranging performance data collected from RM1 presented elsewhere (Fuess et al. 2021a) An inverse pattern was observed when using fresh molasses as the substrate, in which lower RCH4’ values were observed at higher F/M ratios (Table 3), until reaching a marked inhibition of the methanogenic activity in the highest F/M ratio level (3.0 g-COD g−1VS) assessed (Fig. 2c). The carbohydrate-rich composition of Mol (> 80% of the CODs; Table 1) promptly favored the growth of fermentative bacteria, diverting the electron flow, i.e. CODs consumption, from methanogenesis towards cell synthesis. Similar results were observed in previous studies on the biodigestion of highly fermentable substrates, such as sugarcane juice and molasses (Fuess et al. 2021c; Vilela et al. 2021) and glycerol-supplemented sugarcane vinasse (Borges et al. 2022). In all cases, the impairment of the methanogenic activity was not impeditive for achieving relatively high substrate consumption levels, reinforcing the diversion towards cell synthesis. Comparing methane evolution patterns (and the kinetic parameters estimated) for the RM2(fMol) and RM2(Mol) conditions (Fig. 2 and Table 3) supports this hypothesis, because the lowest RCH4 and RCH4’ values were always associated with the non-fermented molasses (for a given F/M ratio). In practical aspects, this discrepancy demonstrated that phase separation increases the capability of anaerobic systems to withstand organic overloading conditions (simulated by high F/M ratios) far beyond the limits achieved in single-stage AD. The experimental arrangement indicated that two-stage schemes withstand at least a threefold higher organic load compared to single-stage ones, because methane production collapsed in condition 3.0-RM2(Mol)—despite the observation of methane production in this case (Fig. 2c), terminal pH values lower than 5.0 were observed, i.e. an evidence of VFA buildup. Hence, starting up a second-stage methanogenic reactor should require lower amounts of sludge and would demand a shorter period compared to single-stage schemes. Shifting the analysis perspective, the comparison between the different biomass sources assessed (in a fixed F/M ratio) indicated that RCH4/RCH4’ values always followed the order RM1 > RM2 > IN (Table 3), suggesting that biomass collected in RM1 was better adapted to the substrate and operation (incubation) conditions than the other inocula. Regardless of the differences observed in the long-term performance of RM1 and RM2, as a direct consequence of different alkalinization strategies, CODs removal efficiency in the feeding zone of both reactors was equivalent, i.e. 74.9% (RM1) and 71.8% (RM2) (Fuess et al. 2021a), indicating that equivalent methane evolution patterns should be observed in MPP tests. Differences in the prevailing microbial communities (also resulting from the different alkalinization approaches) may explain the observed patterns (Section “Microbial community composition”). It is worth highlighting the significant discrepancy in the performance of methanogenesis of IN relative to RM1/RM2, mainly with respect to much longer lag phase periods (Table 3), which resulted from the continuous adaptation of the thermophilic biomass to fermented molasses in RM1 and RM2 for periods of at least 230 d. Besides the microbial differences (Section “Microbial community composition”), the OLR applied in the UASB from which IN was collected (~ 5.0 kg-CODt m−3 d−1) is half of that applied in RM1 and RM2 (10 kg-CODt m−3 d−1) when microbial community samples were obtained, indicating a higher capability to withstand high substrate availability levels in the latter. However, the non-adaptation of the biomass to the incubation conditions was attenuated at high substrate availability: when comparing the different inocula (fMol-fed reactors), the higher the F/M ratio, the lower the variation coefficient calculated for RCH4 (or RCH4’) values, i.e. 30%, 40% and 50% respectively for the F/M ratios of 3.0, 1.0, 0.4 g-COD g−1VS. Kinetics of substrate consumption Differently from methane production, CODs decay followed distinct kinetic patterns, according to both the sludge source and F/M ratio (Fig. 3 and Table 4). Single kinetic decays were observed only when assessing an F/M ratio of 0.4 g-COD g−1VS in biomass samples from RM1 and RM2 (considering both fMol and Mol in the latter). The relative excess sludge amount in such cases favored a prompt substrate consumption, with a wide advantage for the case of RM1 (k1’ = 1.384 × 10–2 h−1 g−1VS; Table 4). Single decay patterns were also observed in the case of IN (F/M ratios of 0.4 and 1.0 g-COD g−1VS); however, only after 106 h of experimental run (Fig. 3a-b), suggesting a progressive adaptation of the inoculum to the incubation conditions. In all remaining cases, substrate decay followed two distinct patterns, either with an initial linear decay (condition 3.0-RM1(fMol)) or with two sequential exponential decays (conditions 1.0-RM2(fMol), 3.0-RM2(fMol), 3.0-IN(fMol) and 1.0-RM2(Mol)) (Table 4). In these conditions, no trend repetition was observed, i.e. while k1’ in the first decay period exceed that of the second in 1.0-RM2(fMol) and 3.0-IN(fMol), the opposite was observed in 3.0-RM2(fMol) and 1.0-RM2(Mol). For a given sludge source (fed with fMol), the higher the F/M ratio, the higher the value of k1’ (Table 4), repeating the behavior previously observed and described for RCH4’ (Table 3). Complementing the hypothesis of higher metabolic efficiency towards methane production, substrate consumption also includes processes other than methanogenesis, such as cell synthesis. Hence, the high k1’ values estimated under lower biomass amounts (high F/M ratios) most likely resulted from both metabolic steps. In fact, relatively high biomass concentrations, i.e. M > F, are recommended when assessing the methanogenic activity of anaerobic sludge sources to minimize the negative effects of excess biomass synthesis over methane production (Aquino et al. 2007; Holliger et al. 2016). The establishment of enhanced cell growth as a significant electron (substrate) sink also justifies the CODs consumption under unfavorable conditions to methanogenesis when using fresh molasses, mainly under high F/M ratios, i.e. 1.0 and 3.0 g-COD g−1VS. These values are much higher than the ones previously recommended for efficient methane production from easily fermentable substrates, i.e. inoculum-to-substrate ratio higher than 4.0 (or F/M ratio lower than 0.25) (Holliger et al. 2016). Revisiting the parallel with reactor startup, efficient substrate conversion rates coupled to short latency periods will only be achieved when sufficient amounts of biomass are available (Haider et al. 2015). Sulfate reduction could have also contributed to the CODs decay, because sulfidogenesis can be associated with the complete oxidation of organic compounds. However, a significant deviation of electrons towards sulfate reduction should only be observed at high sulfate availability (COD/sulfate < 25.0) (Kiyuna et al. 2017), which was not observed for molasses, regardless of the pre-fermentation (COD/sulfate > 60.0; Table 1). Similarly to the case of RCH4’, k1’ values followed the same order for a given F/M ratio, i.e. RM1 > RM2 > IN (Table 4), so that the discrepancy may be understood from an analogous perspective, focusing on microbial characterization aspects (Section “Microbial community composition”). It is worth highlighting that enhanced CODs decay in Mol-fed reactors at lower F/M ratios (≤ 1.0 g-COD g−1VS) was anticipated in comparison with IN (Fig. 3a-b), despite using fermented molasses in the latter. A similar response occurred in the case of methane evolution from fresh molasses at the lowest F/M ratio (0.4 g-COD g−1VS; Fig. 2a), in which the exponential production phase anticipated the one from fermented molasses in reactors inoculated with IN. These results may be explained by the long-term (≥ 230 d) adaptation of IN in the continuous reactors, which attenuated negative effects of enhanced VFA production over methanogens when processing Mol under higher sludge concentrations. Hence, the efficacy of phase separation in high-rate AD systems cannot be assessed separately from the degree of adaptation of the sludge to the desired operating conditions. Breakdown of liquid phase constituents in biodigested molasses Details of the liquid phase constituents at the end of the incubation periods are depicted in Fig. 5, including the terminal pH, the balance between bicarbonate alkalinity and VFA accumulation and the breakdown of the CODs. Different terminal pH values were observed according to the type of substrate, i.e. approximately 8.0 and 7.5 for fermented (regardless of the F/M ratio) and fresh (except for the F/M ratio of 3.0 g-COD g−1VS) molasses (Fig. 5a). While a marked increase relative to the pH of fermented molasses supplied with NaHCO3 prior to incubation (< 7.0) was observed in fMol-fed reactors, initial and terminal pH values were equivalent (~ 7.5) when using fresh molasses (F/M ratios of 0.4 and 1.0 g-COD g−1VS). In the first case, the non-occurrence of enhanced sugar fermentation positively impacted the balance of partial (bicarbonate) alkalinity (Fig. 5b), with values 1.5-to-2.0-fold higher than in reactors fed with fresh molasses (considering equivalent F/M ratios). These patterns (associated with the higher terminal pH values) indicate an excess of NaHCO3 supply in second-stage methanogenic systems, suggesting the possibility to use lower doses without impairing the performance of methanogenesis. In fact, dosing 0.20 g-NaHCO3 g−1COD in RM1 (continuous reactor) did not negatively impact system performance (Fuess et al. 2021a). Conversely, removing NaHCO3 supply triggered the inhibition of methanogenesis in a continuous reactor (also an AnSTBR) fed with fresh molasses, so that a minimum dose of 0.25 g-NaHCO3 g−1COD was defined as the limiting condition to maintain stable performance levels (Oliveira et al. 2020). As a side-effect remark of the experiment, dosing equivalent amounts of NaHCO3 (0.25 g g−1COD) in methanogenic systems fed with fresh and fermented substrates results in excess bicarbonate alkalinity in the latter (up to 200% higher concentrations than in the first). In practical aspects, NaHCO3 dosing in two-stage biodigestion systems have potential to be decreased to economically cost-competitive levels, improving mechanisms to ensure the maintenance of operating stability in industrial scale plants.Fig. 5 Characterization of biodigested molasses at the end of the incubation periods: a pH,b partial alkalinity (PA), c intermediate-to-partial alkalinity (IA/PA) ratio, d concentration of volatile organic acids by titration (VOAtit), e breakdown of VFA (using gas chromatography) and f breakdown of the CODs. Legend: aAfter NaHCO3 dosing, (*) indicates null values The intermediate-to-partial alkalinity (IA/PA) ratio measured at the end of the incubation periods reached values lower than 0.16 in all cases in which methane production was not inhibited (Fig. 5c), indicating much higher participation of bicarbonate-derived alkalinity than the one related to organic acids (despite the aforementioned discrepant patterns). Overall, the lower the F/M ratio, the lower the IA/PA value for a given source of inoculum (Fig. 5c), characterizing a relative accumulation of organic acids at high F/M ratios (VOAtit usually below 500 mg L−1; Fig. 5d). Residual VOAtit concentrations measured in reactors inoculated with biomass from RM1 (426 and 718 mg L−1 in F/M ratios of 1.0 and 3.0 g-COD g−1VS, respectively; Fig. 5d) were higher than those in reactors using RM2-derived sludge (< 283 mg L−1; Fig. 5d) most likely due to differences in the incubation periods. While methane production (CV < 5%; Section “Methane production potential tests”) and the associated CODs decay met the criteria for finalizing the experimental runs within 163 h in the case of RM1, reactors inoculated with sludge from RM2 required longer periods, which varied from 192 h (F/M = 0.4 and 1.0 g-COD g−1VS) to 216 h (F/M = 3.0 g-COD g−1VS). Enhanced acidification (VOAtit = 4500 mg L−1; Fig. 5d) was only observed in condition 3.0-RM2(Mol), in which the inhibition of methanogenesis (Fig. 2c) was associated with terminal pH values lower than 5.0 (Fig. 5a), as well as with the full consumption of the partial alkalinity (Fig. 5b). The distribution of residual VFA concentrations (Fig. 5e) indicated distinct patterns according to the success in the establishment of methanogenesis. Propionate was the prevailing metabolite in conditions in which methane production evolved over time, regardless of differences in the kinetic patterns and type of substrate (Fig. 5e). In these cases, propionate oxidation into acetate and the subsequent oxidation of acetate into CO2 and hydrogen most likely characterized the limiting steps, slowing down methane evolution via the hydrogenotrophic pathway, i.e. the primary methanogenic pathway established in all reactors, as supported by the microbial characterization of the inocula (Section “Microbial community composition”). Differently, acetate buildup was observed in condition 3.0-RM2(Mol), i.e. the only case in which methane production was impaired (Fig. 5e). In addition to the failure in acetate oxidation, which may have also limited the activity of hydrogenotrophic methanogens in this case, the activity of acetoclastic methanogens was equally inefficient to convert the excess acetate available, most likely due to the inhibition by the low pH (Hao et al. 2012). Apart from the accumulation of VFA, the breakdown of the terminal CODs indicated consistent participation of total phenols (PheOH) in the group of non-converted organic matter, with most values within the range of 20 to 40% (Fig. 5f). PheOH usually accounts for a significant fraction of the residual COD in biodigested sugarcane-derived substrates, such as vinasse (Santos et al. 2019) and molasses (Fuess et al. 2021a). The presence of highly recalcitrant melanoidins (from the Maillard reaction between sugars and proteins) and colorants (except caramels) configures the main natural sources of phenols in these substrates (Chandra et al. 2018; Mohana et al. 2009), providing their characteristic dark brown color. It is noteworthy that the participation of PheOH in the CODs increased in all assessed conditions, i.e. from 2.1–2.7% (non-biodigested substrates; Table 1) to (6.2–40.5% (biodigested substrates; Fig. 5e), with the highest terminal values usually observed at excess biomass concentrations (F/M = 0.4 and 1.0 g-COD g−1VS). This pattern most likely resulted from a more intense hydrolytic activity, as previously observed in the dark fermentation of sugarcane vinasse (Piffer et al. 2021). The (partial) breakdown of complex structures (such as those found in melanoidins) implies the release of phenolic compounds to the soluble phase, explaining the high participation of PheOH in the residual CODs. Microbial community composition The composition of the microbial community samples collected at the beginning (inoculum) and at the end of the operation of the continuous methanogenic reactors (RM1 and RM2 in two different sampling points: FDZ and STB) and further used in MPP tests was determined using 16S rRNA gene amplicon sequencing. Prior to detailing the microbial composition of the samples and making inferences about the primary metabolic pathways, PCoA-based comparisons provided bases to understand differences and similarities among all samples from an overall perspective (Fig. 6), considering the interference of: [i] different regions (FDZ and STB) within a given reactor, [ii] different reactors (RM1 and RM2), [iii] the specialization of the sludge (IN) after long-term (≥ 230 d) operating periods, and [iv] the impacts of sample storage. It is worth highlighting that analyzing the impacts of sample storage was simply a parallel investigation of the study, in an effort to understand potential failures in MPP tests (which were not observed). PCoA results were interpreted considering both the total microbial community (Archaea and Bacteria domains; Fig. 6a) and exclusively the archaeal community (Fig. 6b).Fig. 6 Principal coordinates analysis (PCoA, UniFrac) plot for a total community and b archaeal community. Legend: samples related to the inoculum (filled circle), RM1 (filled triangle) and RM2 (filled square). “ps” indicates post-storage Considering the total microbial community, while little differentiation was observed between different compartments of a given reactor (regardless of the storage in the case of RM2-FDZ), a marked distance was identified between RM1- and RM2-derived samples (Fig. 6a), which were grouped separately. Although the operation of both reactors was finalized under equivalent conditions of OLR (10.0 kg-CODt m−3 d−1), HRT (24.0 h) and alkalinization (NaHCO3 dosing), RM2 was subjected to events of enhanced VFA accumulation when dosing NaOH in previous steps of the operation (Fuess et al. 2021a), which may explain the observed discrepancy. Total microbial communities from both reactors were also very different from IN, considering particularities of each AD system in which samples were collected (IN vs. RM1/RM2): reactor type (UASB vs. AnSTBR), substrate type (vinasse vs. fermented molasses) and OLR (5.0 vs. 10.0 kg-CODt m−3 d−1). In the case of IN, the long-term storage (14 months) impacted the microbial composition to a higher extent than in the case of RM2-FDZ. Limiting the analysis to the archaeal community (Fig. 6b), samples collected from RM1 and RM2 were very similar, regardless of the compartment. This particular result confirms that variations in the bacterial community of RM2 triggered the discrepancy previously observed in the total microbial communities of both RM1 and RM2 (Fig. 6a). Marked difference was still observed between samples from RM1/RM2 and IN. Interestingly, long-term storage impacted the archaeal community to a much higher extent than the different growth conditions (suspended vs. attached) along the reactors, using the differences between samples RM2-FDZ and RM2-FDZps as the references. Hence, the relative similarity between microbial communities from different compartments of given reactor (regardless of considering only Archaea or both Archaea and Bacteria domains) answers one of the questions raised in Introduction, indicating that a more homogeneous microbial distribution is expected to occur in the biodigestion (methanogenesis) of pre-fermented substrates. However, the unbalanced conversion of organic matter observed in the continuous reactors, i.e. > 70% in the FDZ of both systems (Fuess et al. 2021a), leads to further questions: [i] How metabolically active are the microbial communities located in the bed region? [ii] Could the fixed bed simply be a “cell retention barrier” in the proposed reactor configuration? Details of the microbial characterization of the biomass samples at the phylum and genus levels are depicted in Fig. 7, also considering the total microbial community (Archaea and Bacteria domains; Fig. 7a-b) and an isolated analysis of the Archaea domain (Fig. 7c-d). Additionally, heat maps for ASV and genera are presented in the Supplementary data section for further reference. Overall, the Firmicutes phylum prevailed in all samples collected from RM1 and RM2 (Fig. 7a), characterizing a marked different pattern compared to IN and INps. While the phyla Campylobacterota, Firmicutes and Proteobacteria were identified relatively in equivalent abundances in IN, the Proteobacteria phylum prevailed in INps (Fig. 7a). The phylum Euryarchaeota, in which most methanogens are included (Lyu and Liu 2018), was the primary archaeal group identified in all samples (Fig. 7c; except for the case of INps, in which the phylum Halobacterota was also relevant). The Methanothermobacter genus was the most abundant methanogen identified in all samples collected from RM1 and RM2 (relative abundance—RA = 56.7–79.7% among Archaea; Fig. 7d), regardless of the growth condition, i.e. suspended (FDZ) or attached (STB). This result, in association with the identification of the Methanosaeta genus exclusively in samples from RM1 and RM2 (RA = 7.8–25.5% among Archaea; Fig. 7d), largely explain the similarity observed for archaeal communities (pre-storage) from both continuous reactors (Fig. 6b). The storage-dependent difference observed for samples RM2-FDZ and RM2-FDZps resulted from shifts in minor ASV associated with each genus identified. Methanothermobacter also prevailed in the thermophilic sludge (IN) used in the inoculation of RM1 and RM2 (RA = 64.7% among Archaea) and was followed by the Methanoculleus genus as the second most abundant archaeal group (RA = 18.3%) (Fig. 7d). However, the latter (RA = 48.9%) surpassed the first (RA = 41.3%) after the long-term storage of IN (INps; Fig. 7d), which effectively characterized the primary distribution of methanogens during MPP tests. The significant presence of the Methanoculleus genus in both IN and INps samples (and the absence of the Methanosaeta genus) explains their difference relative to RM1 and RM2 (Fig. 6b), whilst the increase in the RA for Methanoculleus in INps explains the distance relative to IN. Methanosarcina (RA < 10.5%) and Bathyarchaeia (RA < 9.0%) genera were also identified in all biomass samples (Fig. 7d), but at lower RA levels compared to the previous groups.Fig. 7 Taxonomic distribution according to the 16S rRNA gene amplicon sequencing analysis of the total microbial community at phylum and genus level (a, b), and archaeal community at phylum and genus level (c, d). “ps” indicates post-storage Interestingly, the most abundant methanogens identified in all samples, i.e. Methanothermobacter (Hao et al. 2012; Cheng et al. 2011; Wasserfallen et al. 2000) and Methanoculleus (Dyksma et al. 2020; Manzoor et al. 2016; Shigematsu et al. 2004) genera, grow autotrophically on H2 and CO2, which may sound contradictory when considering two-stage AD schemes. Nearly 90% of the total carbohydrates found in fresh molasses were converted during dark fermentation (Fuess et al. 2021b) prior to feeding RM1 and RM2, markedly eliminating the prompt provision of both H2 and CO2 by acidogenic populations in the methanogenic units. However, the predominance of both genera in the inoculum (combined RA > 80% among Archaea; Fig. 7d) associated with the relatively low availability of acetate (4.6% of the CODs; Table 1) in fermented molasses, i.e. much lower levels than those of lactate (52.9%) and butyrate (17.1%) (Table 1), required the establishment of a strong association between acetogens, syntrophic acetate-oxidizing bacteria (SAOB) and hydrogenotrophic methanogens (HM) in both RM1 and RM2. In particular, the association between SAOB and HM has been systematically pointed to replace acetoclastic methanogenesis under thermophilic conditions (Hao et al. 2012; Cheng et al. 2011; Dyksma et al. 2020), which supports the occurrence of the metabolic sequence proposed. Compared to acetoclastic methanogenesis, methane production from the association between SAOB and HM is thermodynamically favored at high temperatures (Dolfing 2014). In addition, previous investigations indicated the prevalence of the latter pathway at relatively low acetate availability (Shigematsu et al. 2004; Petersen and Ahring 1991), which is in accordance with the metabolite distribution profile in fermented molasses (Table 1). Differently from the participation of methanogens, notably characterized by a few representative genera (Fig. 7d), a higher diversity of acetogenic and acetate-oxidizing groups at much lower abundance levels was identified. The oxidation of fermentation metabolites into acetate was most likely carried out by the genera Pelotomaculum (RA = 0.9–6.6%, except in IN and INps), Syntrophothermus (RA = 0.6–3.8%, except in IN and INps), Cloacimonadacea W5 (RA = 1.3–5.9%, except in IN and INps), Syntrophomonas (RA = 0.9–4.6%, except in IN and INsp), and Thermodesulfovibrio (RA = 0.3–4.3%, all samples). Butyrate oxidation into acetate was previously associated with the Syntrophothermus (Sekiguchi et al. 2000) and Syntrophomonas (McInerney et al. 1981) genera, whilst both Pelotomaculum (Imachi et al. 2007) and Cloacimonadaceae W5 (Dyksma and Gallert 2019) genera were involved in propionate oxidation. In all cases, syntrophic associations with methanogens were obligate to maintain low acetate levels, characterizing a different aspect of the cases reported in this study, because SAOB most likely consumed most of the available acetate. Interestingly, propionate availability in fermented molasses was much lower than all other fermentation metabolites (0.5% of the CODs; Table 1), so that an intermediate step releasing propionate was most likely established prior to acetogenesis. In particular lactate fermentation into acetate and propionate by species belonging to the Clostridium and Veillonella genera was identified as a relevant pathway in methanogenic environments (Seeliger et al. 2002; Stams et al. 1998), potentially fitting the proposed intermediate step. An ASV associated with the Veillonellales-Senomonadales order was identified primarily in biomass samples collected from the FDZ of both RM1 and RM2 (RA = 6.1–13.5%), which may support this hypothesis. Alternatively, Thermodesulfovibrio may have mediated the syntrophic degradation of lactate, which has been previously reported to occur in both the absence of sulfate and co-culture with hydrogenotrophic methanogens (Sekiguchi et al. 2008). The low sulfate availability in molasses (COD/sulfate > 60.0; Table 1) does not support the long-term persistence of such a sulfate-reducing group in the continuous reactors, suggesting the establishment of an alternative metabolic pathway to maintain cell growth. Following the SAOB-HM pathway, genera responsible for mediating syntrophic acetate oxidation included Thermoacetogenium (RA = 1.4–2.3%, RM1-FDZ and RM1-STB), Mesotoga (RA = 3.4–7.8%, all samples) and Pseudothermotoga (RA = 0.2–2.1%, except in IN and INps). The Thermoacetogenium genus has been associated with high metabolic flexibility, according to the environmental conditions offered (Hattori et al. 2000). On one hand, acetate oxidation has been reported to occur in co-culture with hydrogenotrophic microorganisms or associated with sulfate reduction in pure culture. On the other hand, acetate production from alcohol fermentation or autotrophically from H2 and CO2 (homoacetogenesis) has also been observed. Acetate oxidation by Pseudothermotoga has been observed only in the presence of thiosulfate or methanogens (Balk et al. 2002), whilst in the case of Mesotoga the acetate-oxidizing pathway is still uncharacterized (Nobu et al. 2015). However, investigations using labelled carbon suggested that the latter syntrophically oxidize acetate only in very low acetate concentrations (Wang et al. 2019). Concurrently to SAOB, Methanosaeta most likely mediated secondary acetate utilization in methanogenesis via the acetoclastic pathway. Differently from Methanosarcina, which is able to mediate acetoclastic methanogenesis under high acetate availability (Petersen and Ahring 1991; Hori et al. 2006), the activity of Methanosaeta is favored under low acetate concentrations (Petersen and Ahring 1991; Karakashev et al. 2005; Suárez et al. 2018). Hence, efficient consumption of acetate by SAOB (maintained by the high activity of HM) made only a reduced proportion of acetate available for Methanosaeta, characterizing the secondary participation of this group in methane evolution. Interestingly, the highest RA levels observed for Methanosaeta were identified in the STB of both reactors, mainly in the case of RM2 (25.5%; Fig. 7d). Because of VFA accumulation events prior to the finalization of the operation, acetate concentrations higher than 1000 mg-COD L−1 were measured in RM2 (Fuess et al. 2021a), suggesting higher participation of Methanosaeta in methane evolution. From a metabolic perspective, the SAOB-HM pathway was not able to efficiently metabolize excess acetate, requiring the establishment of parallel acetate-consuming pathways. High RA for the Methanosarcina genus was also identified in RM2 (7.4%, FDZ; Fig. 7d), which supports the hypothesis of favoring parallel acetate-consuming pathways. This group may also have participated as a HM, because it includes species capable of growing using any of the known methanogenic pathways, i.e. acetoclastic, hydrogenotrophic, carboxydotrophic (from carbon monoxide), methylotrophic (from methanol, methylamines and methylsulfides) and methyl respiration (methylated compounds + H2) (Buan et al. 2011). The microbial characterization also provided bases to understand the behavior of methane evolution in MPP tests. The long lag phase periods observed in reactors inoculated with INps (Fig. 2a–c) most likely resulted from non-identifying genera previously pointed out as key-players in acetogenesis (e.g. Pelotomaculum, Syntrophothermus and Cloacimonadaceae W5) and syntrophic acetate oxidation (e.g. Mesotoga and Pseudothermotoga) in the continuous reactors. Null RA values were equally identified for the Methanosaeta genus (Fig. 7d) in INps, suggesting an even stronger dependence of methanogenesis on the SAOB-HM pathway. This hypothesis is corroborated by the delay periods observed in CODs decay when assessing INps (Fig. 3a-c), because an effective participation of acetoclastic methanogenesis would promptly trigger CODs conversion. Because the Thermodesulfovibrio genus was present in INps (at very low initial RA, i.e. 0.3%), this group may have had an important participation in supplying the SAOB-HM pathway on the course of the incubation. Considering RM1- and RM2-derived samples, the minor availability of biomass samples of RM1-FDZ was impeditive to effectively comparing the sludge samples used in MPP tests (post-storage). However, the comparison between RM1-FDZ and RM2-FDZ (both collected during the disassembling of the continuous reactors) indicated higher RA values for key acetogenic (Cloacimonadacea W5, Pelotomaculum, Syntrophothermus and Syntrophomonas), SAOB (Mesotoga, Thermoacetogenium and Pseudothermotoga) and methanogenic (Methanothermobacter) genera in the first sample, explaining the better performances regarding methane evolution when using sludge from RM1. Moreover, the differences in MPP tests most likely did not reach higher levels because the participation of important groups increased in RM2-FDZps (relative to RM2-FDZ), such as the genera Pelotomaculum, Syntrophothermus, Methanothermobacter and Methanosaeta. In any case, once the association between acetogens, SAOB and HM was efficiently established (regardless of the sludge source), both methane evolution and CODs decay proceeded more efficiently than in previous investigations with both fresh and fermented molasses under mesophilic temperature (30 ºC) (Fuess et al., 2020). In these cases, the dependence of acetoclastic methanogenesis on the activity of acetogens markedly slowed down methane evolution, which occurred at linear (and not exponential) rates in periods longer than 100 h during the incubation (Fuess et al. 2020). Hence, whenever possible, using thermophilic conditions to intentionally select HM may also characterize an essential approach to increase the efficiency of methane production in two-stage AD, because HM present much lower doubling times (a few hours) compared to acetoclastic groups (a few days) (Mosey 1983). The capability of phase separation to increase the robustness of methane evolution relies more on this aspect than on “simply” eliminating enhanced fermentation from the methanogenic unit. Finally, sugar fermentation was most likely carried out by the genera Acetomicrobium (Hania et al. 2016), Anaerolinea (Sekiguchi et al. 2003), Lentimicrobium (Sun et al. 2016), Fervidobacterium (Cai et al. 2007) and Caldicoprobacter (Yokoyama et al. 2010) when using fresh molasses in RM2(Mol) tests, supplying sequential acetogenic and acetate-oxidizing groups. Conclusions The primary conclusions drawn from this study include:Phase separation under thermophilic conditions is the best approach to achieve efficient methane production from sugar-rich substrates. In addition to the consolidated idea of minimizing stressful conditions to methanogens by separating enhanced substrate fermentation, the use of high temperatures favor the participation of hydrogenotrophic methanogenesis. In practical aspects, methane evolution will depend on microbial groups (e.g. Methanothermobacter and Methanoculleus genera) that grow faster and are less susceptible to low pH values compared to acetoclastic ones (e.g. Methanosaeta genus). Following the experimental results, the capability of second-stage methanogenic systems to withstand organic loads can be increased by at least threefold compared to single-stage schemes; The microbial community distribution tends to be less stratified in the methanogenic unit of the two-stage biodigestion system, regardless of the differences in both the substrate availability (once the F/M ratio decreases as the liquid flows through the reactor) and the conditions provided for the cell growth (suspended in the bulk liquid or attached to the fixed media). In particular, homogeneity in the distribution of syntrophic acetate-oxidizing bacteria (e.g. Mesotoga and Thermoacetogenium) may be sine qua non for achieving all benefits previously associated with thermophilic two-stage biodigestion (conclusion no. [i]); and, Further studies are still required to better understand the role of microbial communities attached to the fixed bed in such methanogenic reactors. In particular, the scale-up of the AnSTBR may be imperative to unravel the distribution of microbial communities in the different compartments of the system. The concentration of biomass in the feeding zone of bench-scale reactors tends to be proportionally high compared to that of the bed region, triggering a sharp substrate decay which limits defining the effective participation of the attached biomass in the overall substrate conversion. Supplementary Information Below is the link to the electronic supplementary material.Supplementary file1 (PDF 1741 KB) Acknowledgements This work was supported by the São Paulo Research Foundation (FAPESP) [grant numbers 2017/00080-5, 2015/50684-9, 2015/06246-7 and 2014/50279-4]; and the Coordination for the Improvement of Higher Education Personnel via the Academic Excellence Program (PROEX-CAPES). Declarations Conflict of interest The authors declare that they have no known competing interests that could have appeared to influence the work reported in this paper. Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References Adorno MAT Hirasawa JS Varesche MBA Development and validation of two methods to quantify volatile acids (C2–C6) by GC/FID: headspace (automatic and manual) and liquid-liquid extraction (LLE) Am J Anal Chem 2014 5 406 414 10.4236/ajac.2014.57049 Andersen KS, Kirkegaard RH, Karst SM, Albertsen M (2018) ampvis2: an R package to analyse and visualise 16S rRNA amplicon data. bioRxiv 299537. 10.1101/299537 Anzola-Rojas MP Fonseca SG Silva CC Oliveira VM Zaiat M The use of the carbon/nitrogen ratio and specific organic loading rate as tools for improving biohydrogen production in fixed-bed reactors Biotechnol Rep 2015 5 46 54 10.1016/j.btre.2014.10.010 APHA, Awwa, WEF, Standard methods for the examination of water and wastewater 2012 Washington, DC APHA Aquino SF Chernicharo CAL Foresti E Santos MLF Monteggia LO Methodologies for determining the specific methanogenic activity (SMA) in anaerobic sludges Eng Sanit Ambient 2007 12 2 192 201 10.1590/S1413-41522007000200010[inPortuguese] Aquino S Fuess LT Pires EC Media arrangement impacts cell growth in anaerobic fixed-bed reactors treating sugarcane vinasse: structured vs. randomic biomass immobilization Bioresour Technol 2017 235 219 228 10.1016/j.biortech.2017.03.120 28365350 Balk M, Weijma J, Stams AJM (2002) Thermotoga lettingae sp. nov., a novel thermophilic, methanol-degrading bacterium isolated from a thermophilic anaerobic reactor. Int J Syst Evol Microbiol 52(4):1361–1368. 10.1099/00207713-52-4-1361 Barros VG Duda RM Olveira RA Biomethane production from vinasse in upflow anaerobic sludge blanket reactors inoculated with granular sludge Braz J Microbiol 2016 47 3 628 639 10.1016/j.bjm.2016.04.021 27289246 Blanco VMC Fuess LT Zaiat M Calcium dosing for the simultaneous control of biomass retention and the enhancement of fermentative biohydrogen production in an innovative fixed-film bioreactor Int J Hydrogen Energy 2017 42 12181 12196 10.1016/j.ijhydene.2017.02.180 Bolyen E Rideout JR Dillon MR Bokulich NA Abnet CC Al-Ghalith GA Reproducible, interactive, scalable and extensible microbiome data science using QIIME 2 Nat Biotechnol 2019 37 852 857 10.1038/s41587-019-0209-9 31341288 Borges AV Fuess LT Takeda PY Alves I Dias MES Damianovic MHRZ Co-digestion of biofuel by-products: Enhanced biofilm formation maintains high organic matter removal when methanogenesis fails J Environ Manage 2022 310 114768 10.1016/j.jenvman.2022.114768 35220101 Buan N Kulkarni G Metcalf W Genetic methods for Methanosarcina species Methods Enzymol 2011 494 23 42 10.1016/B978-0-12-385112-3.00002-0 21402208 Buchanan ID Nicell JA Model development for horseradish peroxidase catalyzed removal of aqueous phenol Biotechnol Bioeng 1997 54 3 251 261 10.1002/(SICI)1097-0290(19970505)54:3<251::AID-BIT6>3.0.CO;2-E 18634091 Cai J, Wang Y, Liu D, Zeng Y, Xue Y, Ma Y, Feng Y (2007) Fervidobacterium changbaicum sp. nov., a novel thermophilic anaerobic bacterium isolated from a hot spring of the Changbai Mountains, China. Int J Syst Evol Microbiol 57(10):2333–2336. 10.1099/ijs.0.64758-0 Callahan BJ McMurdie PJ Rosen MJ Han AW Johnson AJ Holmes SP DADA2: High-resolution sample inference from Illumina amplicon data Nat Methods 2016 13 581 583 10.1038/nmeth.3869 27214047 Camiloti PR Mockaitis G Rodrigues JAD Damianovic MHRZ Foresti E Zaiat M Innovative anaerobic bioreactor with fixed-structured bed (ABFSB) for simultaneous sulfate reduction and organic matter removal J Chem Technol Biotechnol 2014 89 7 1044 1050 10.1002/jctb.4199 Chan YJ Chong MF Law CL Hassell DG A review on anaerobic-aerobic treatment of industrial and municipal wastewater Chem Eng J 2009 155 1–2 1 18 10.1016/j.cej.2009.06.041 Chandra R, Kumar V, Tripathi S (2018) Evaluation of molasses-melanoidin decolourisation by potential bacterial consortium discharged in distillery effluent. 3 Biotech 8:187. 10.1007/s13205-018-1205-3 Cheng L, Dai L, Li X, Zhang H, Lu Y (2011) Isolation and characterization of Methanothermobacter crinale sp. nov., a novel hydrogenotrophic methanogen from the Shengli oil field. Appl Environ Microbiol 77(15):5212–5219. 10.1128/AEM.00210-11 Chernicharo CAL Anaerobic reactors IWA Publishing, London 2007 10.2166/9781780402116 Dolfing J Thermodynamic constraints on syntrophic acetate oxidation Appl Environ Microbial 2014 80 4 1539 1541 10.1128/AEM.03312-13 Dubois M Gilles KA Hamilton JK Rebers PA Smith F Colorimetric methods for determination of sugar and related substance Anal Chem 1956 28 3 350 356 10.1021/ac60111a017 Dyksma S Gallert C Candidatus Syntrophosphera thermopropionivorans: a novel player in syntrophic propionate oxidation during anaerobic digestion Environ Microbiol Rep 2019 11 4 558 570 10.1111/1758-2229.12759 30985964 Dyksma S Jansen L Gallert C Syntrophic acetate oxidation replaces acetoclastic methanogenesis during thermophilic digestion of biowaste Microbiome 2020 8 105 10.1186/s40168-020-00862-5 32620171 Fuess LT, Zaiat M, Nascimento CAO (2020) Molasses vs. juice: Maximizing biohydrogen production in sugarcane biorefineries to diversify renewable energy generation. J Water Process Eng 37:101534. 10.1016/j.jwpe.2020.101534 Fuess LT Zaiat M Nascimento CAO Thermophilic biodigestion of fermented sugarcane molasses in high-rate structured-bed reactors: Alkalinization strategies define the operating limits Energy Convers Manage 2021 239 114203 10.1016/j.enconman.2021.114203 Fuess LT Fuentes L Bovio-Winkler P Eng F Etchebehere C Zaiat M Nascimento CAO Full details on continuous biohydrogen production from sugarcane molasses are unraveled: Performance optimization, self-regulation, metabolic correlations and quanti-qualitative biomass characterization Chem Eng J 2021 414 128934 10.1016/j.cej.2021.128934 Fuess LT Cruz RBCM Zaiat M Nascimento CAO Diversifying the portfolio of sugarcane biorefineries: Anaerobic digestion as the core process for enhanced resource recovery Renew Sustain Energy Rev 2021 147 111246 10.1016/j.rser.2021.111246 Hafez H Nakhla G El. Naggar MH, Elbeshbishy E, Baghchehsaraee B, Effect of organic loading on a novel hydrogen bioreactor Int J Hydrogen Energy 2010 35 1 81 92 10.1016/j.ijhydene.2009.10.051 Haider MR Zeshan YS Malik RN Visvanathan C Effect of mixing ratio of food waste and rice husk co-digestion and substrate to inoculum ratio on biogas production Bioresour Technol 2015 190 451 457 10.1016/j.biortech.2015.02.105 25818922 Hania WB, Bouanane-Darenfed A, Cayol JL, Ollivier B, Fardeau ML (2016) Reclassification of Anaerobaculum mobile, Anaerobaculum thermoterrenum, Anaerobaculum hydrogeniformans as Acetomicrobium mobile comb. nov., Acetomicrobium thermoterrenum comb. nov. and Acetomicrobium hydrogeniformans comb. nov., respectively, and emendation of the genus Acetomicrobium. Int J Syst Evol Microbiol 66(3):1506–1509. 10.1099/ijsem.0.000910 Hao LP Lü F Li L Shao LM He PJ Shift of pathways during initiation of thermophilic methanogenesis at different initial pH Bioresour Technol 2012 126 418 424 10.1016/j.biortech.2011.12.072 22227145 Hattori S, Kamagata Y, Hanada S, Shoun H (2000) Thermoacetogenium phaeum gen. nov., sp. nov., a strictly anaerobic, thermophilic, syntrophic acetate-oxidizing bacterium. Int J Syst Evol Microbiol 50(4):1601–1609. 10.1099/00207713-50-4-1601 Holliger C Alves M Andrade D Angelidaki I Astals S Baier U Towards a standardization of biomethane potential tests Water Sci Technol 2016 74 11 2515 2522 10.2166/wst.2016.336 27973356 Hori T Haruta S Ueno Y Ishii M Igarashi Y Dynamic transition of a methanogenic population in response to the concentration of volatile fatty acids in a thermophilic anaerobic digester Appl Environ Microbiol 2006 72 2 1623 1630 10.1128/AEM.72.2.1623-1630.2006 16461718 Imachi H, Sakai S, Ohashi A, Harada H, Hanada S, Kagamata Y, Sekiguchi Y (2007) Pelotomaculum propionicum sp. nov., an anaerobic, mesophilic, obligately syntrophic, propionate-oxidizing bacterium. Int J Syst Evol Microbiol 57(7):1487–1492. 10.1099/ijs.0.64925-0 Kapp H (1984) Schlammfaulung mit hohem Feststoffgehalt. Stuttgarter Berichte zur Siedlungswasserwirtschaft, Band 86:300, Oldenbourg Verlag, München, Munich, Germany (Sludge with a High Solids Content, Stuttgart reports for Urban Water Management). Karakashev D Batstone DJ Angelidaki I Influence of environmental conditions on methanogenic compositions in anaerobic biogas reactors Appl Environ Microbiol 2005 71 1 331 338 10.1128/AEM.71.1.331-338.2005 15640206 Kiyuna LSM Fuess LT Zaiat M Unraveling the influence of the COD/sulfate ratio on organic matter removal and methane production from the biodigestion of sugarcane vinasse Bioresour Technol 2017 232 103 112 10.1016/j.biortech.2017.02.028 28214696 Lebrero R Toledo-Cervantes A Muñoz R Del Nery V Foresti E Biogas upgrading from vinasse digesters: a comparison between an anoxic biotrickling filter and an algal-bacterial photobioreactor J Chem Technol Biotechnol 2016 91 9 2488 2495 10.1002/jctb.4843 Lyu Z, Liu Y (2018) In: Stams AJM, Sousa D (eds) Biogenesis of Hydrocarbons. Handbook of Hydrocarbon and Lipid Microbiology, Springer, Cham, p. 1–59. 10.1007/978-3-319-53114-4_5-1 Manzoor S Schnürer A Bongcam-Rudloff E Müller B Complete genome sequence of Methanoculleus bourgensis strain MAB1, the syntrophic partner of mesophilic acetate-oxidising bacteria (SAOB) Stand Genomic Sci 2016 11 80 10.1186/s40793-016-0199-x 27777650 McInerney MJ, Bryant MP, Hespell RB, Costerton JW (1981) Syntrophomonas wolfei gen. nov. sp. nov., an anaerobic syntrophic, fatty acid-oxidizing bacterium. Appl Environ Microbiol 41(4):1029–1039. 10.1128/aem.41.4.1029-1039.1981 McMurdie PJ Holmes S phyloseq: An R package for reproducible interactive analysis and graphics of microbiome census data PLoS ONE 2013 8 4 e61217 10.1371/journal.pone.0061217 23630581 Mockaitis G Pantoja JLR Rodrigues JAD Foresti E Zaiat M Continuous anaerobic bioreactor with a fixed-structure bed (ABFSB) for wastewater treatment with low solids and low applied loading content Bioprocess Biosyst Eng 2014 37 1361 1368 10.1007/s00449-013-1108-y 24390575 Mohana S Acharya BK Madamwar D Distillery spent wash: Treatment technologies and potential applications J Hazard Mater 2009 163 1 12 25 10.1016/j.jhazmat.2008.06.079 18675513 Moletta R Winery and distillery wastewater treatment by anaerobic digestion Water Sci Technol 2005 51 1 137 144 10.2166/wst.2005.0017 Mosey FE Mathematical modelling of the anaerobic digestion process: Regulatory mechanisms for the formation of short-chain volatile acids from glucose Water Sci Technol 1983 15 8–9 209 232 10.2166/wst.1983.0168 Nobu MK Narihiro T Rinke C Kamagata Y Tringe SG Woyke T Liu WT Microbial dark matter ecogenomics reveals complex synergistic networks in a methanogenic bioreactor ISME J 2015 9 1710 1722 10.1038/ismej.2014.256 25615435 Oliveira CA Fuess LT Soares LA Damianovic MHRZ Thermophilic biomethanation of sugarcane molasses comparing single and two-stage systems: Process performance and energetic potential Bioresour Technol Rep 2020 12 100590 10.1016/j.biteb.2020.100590 Petersen SP Ahring BK Acetate oxidation in a thermophilic anaerobic sewage-sludge digestor: the importance of non-aceticlastic methanogenesis from acetate FEMS Microbiol Ecol 1991 86 2 149 158 10.1111/j.1574-6968.1991.tb04804.x Piffer MA Zaiat M Nascimento CAO Fuess LT Dynamics of sulfate reduction in the thermophilic dark fermentation of sugarcane vinasse: a biohydrogen-independent approach targeting enhanced bioenergy production J Environ Chem Eng 2021 9 105956 10.1016/j.jece.2021.105956 Piffer MA Oliveira CA Bovio-Winkler P Eng F Etchebehere C Zaiat M Nascimento CAO Fuess LT Sulfate- and pH driven metabolic flexibility in sugarcane vinasse dark fermentation stimulates biohydrogen evolution, sulfidogenesis or homoacetogenesis Int J Hydrogen Energy 2022 47 73 31202 31222 10.1016/j.ijhydene.2022.07.039 Quast C Pruesse E Yilmaz P Gerken J Schweer T Yarza P Peplies J Glöckner FO The SILVA ribosomal RNA gene database project: improved data processing and web-based tools Nucleic Acids Res 2013 41 D1 D590 D596 10.1093/nar/gks1219 23193283 Ripley LE Boyle WC Converse JC Improved alkalimetric monitoring for anaerobic digestion of high-strength wastes J Water Pollut Control Fed 1986 58 5 406 411 Rodrigues JAD Ratusznei SM Camargo EFM Zaiat M Influence of agitation rate on the performance of an anaerobic sequencing batch reactor containing granulated biomass treating low-strength wastewater Adv Environ Res 2003 7 405 410 10.1016/S1093-0191(02)00008-4 Santos PS Zaiat M Nascimento CAO Fuess LT Does sugarcane vinasse composition variability affect the bioenergy yield in anaerobic systems? A dual kinetic-energetic assessment J Clean Prod 2019 240 118005 10.1016/j.jclepro.2019.118005 Seeliger S Janssen PH Schink B Energetics and kinetics of lactate fermentation to acetate and propionate via methylmalonyl-CoA or acrylyl-CoA FEMS Microbiol Lett 2002 211 1 65 70 10.1111/j.1574-6968.2002.tb11204.x 12052552 Sekiguchi Y, Kamagata Y, Nakamura K, Ohashi A, Harada H (2000) Syntrophothermus lipocalidus gen. nov., sp. nov., a novel thermophilic, syntrophic, fatty-acid-oxidizing anaerobe which utilizes isobutyrate. Int J Syst Evol Microbiol 50(2):771–779. 10.1099/00207713-50-2-771 Sekiguchi Y, Yamada T, Hanada S, Ohashi A, Harada H, Kamagata Y (2003) Anaerolinea thermophila gen. nov., sp. nov. and Caldilinea aerophila gen. nov., sp. nov., novel filamentous thermophiles that represent a previously uncultured lineage of the domain Bacteria at the subphylum level. Int J Syst Evol Microbiol 53(6):1843–1851. 10.1099/ijs.0.02699-0 Sekiguchi Y, Muramatsu M, Imachi H, Narihiro T, Ohashi A, Harada H, Hanada S, Kagamata Y (2008) Thermodesulfovibrio aggregans sp. nov. and Thermodesulfovibrio thiophilus sp. nov., anaerobic, thermophilic, sulfate-reducing bacteria isolated from thermophilic methanogenic sludge, and emended description of the genus Thermodesulfovibrio. Int J Syst Evol Microbiol 58(11):2541–2548. 10.1099/ijs.0.2008/000893-0 Shigematsu T Tang Y Kobayashi T Kawaguchi H Morimura S Kida K Effect of dilution rate on metabolic pathway shift between aceticlastic and nonaceticlastic methanogenesis in chemist cultivation Appl Environ Microbiol 2004 70 7 4048 4052 10.1128/AEM.70.7.4048-4052.2004 15240282 Stams AJM Dijkema C Plugge CM Lens P Contribution of 13C-NMR spectroscopy to the elucidation of pathways of propionate formation and degradation in methanogenic environments Biodegradation 1998 9 6 463 473 10.1023/A:1008342130938 Suárez WAB Vantini JS Duda RM Giachetto PF Cintra LC Ferro MIT Oliveira RA Predominance of syntrophic bacteria, Methanosaeta and Methanoculleus in a two-stage up-flow anaerobic sludge blanket reactor treating coffee processing wastewater at high organic loading rate Bioresour Technol 2018 268 158 168 10.1016/j.biortech.2018.06.091 30077172 Sun L, Toyonaga M, Ohashi A, Tourlousse DM, Matsuura N, Meng XY, Tamaki H, Hanada S, Cruz R, Yamaguchi T, Sekiguchi Y (2016) Lentimicrobium saccharophilum gen. nov., sp. nov., a strictly anaerobic bacterium representing a new family in the phylum Bacteroidetes, and proposal of Lentimicrobiaceae fam. nov. Int J Syst Evol Microbiol 66(7):2635–2642. 10.1099/ijsem.0.001103 Taylor KACC A simple colorimetric assay for muramic acid and lactic acid Appl Biochem Biotechnol 1996 56 1 49 58 10.1007/BF02787869 van Lier JB van der Zee FP Frijters CTMJ Ersahin ME Celebrating 40 years anaerobic sludge bed reactors for industrial wastewater treatment Rev Environ Sci Biotechnol 2015 14 681 702 10.1007/s11157-015-9375-5 Vilela RS Fuess LT Saia FT Silveira CRM Oliveira CA Andrade PA Langenhoff A van der Zaan B Cop F Gregoracci GB Damianovic MHRZ Biofuel production from sugarcane molasses in thermophilic anaerobic structured-bed reactors Renew Sustain Energy Rev 2021 144 110974 10.1016/j.rser.2021.110974 Wang HZ Lv XM Yi Y Zheng D Gou M Nie Y Hu B Nobu MK Narihiro T Tang YQ Using DNA-based stable isotope probing to reveal novel propionate- and acetate-oxidizing bacteria in propionate-fed mesophilic anaerobic chemostats Sci Rep 2019 9 17396 10.1038/s41598-019-53849-0 31758023 Wasserfallen A, Nölling J, Pfister P, Reeve J, Macario EC (2000) Phylogenetic analysis of 18 thermophilic Methanobacterium isolates supports the proposals to create a new genus, Methanothermobacter gen. nov., and to reclassify several isolates in three species, Methanothermobacter thermautotrophicus comb. nov., Methanothermobacter wolfeii comb. nov., and Methanothermobacter marburgensis sp. nov. Int J Syst Evol Microbiol 50(1):43–53. 10.1099/00207713-50-1-43 Wickham H ggplot2: Elegant Graphics for Data Analysis 2016 New York Springer-Verlag Yokoyama H, Wagner ID, Wiegel J (2010) Caldicoprobacter oshimai gen. nov., sp. nov., an anaerobic, xylanolytic, extremely thermophilic bacterium isolated from sheep faeces, and proposal of Caldicoprobacteraceae fam. nov. Int J Syst Evol Microbiol 60(1):67–71. 10.1099/ijs.0.011379-0 Zwietering MH, Jongenburger I, Rombouts FM, van’t Riet K (1990) Modeling of the bacterial growth curve. Appl Environ Microbiol 56(6):1875–1881. 10.1128/aem.56.6.1875-1881.1990

0

PMC9753886

NO-CC CODE

2022-12-16 23:26:21

no

Braz. J. Chem. Eng.. 2022 Dec 15;:1-21

utf-8

oa_other

==== Front Vegetos Vegetos Vegetos (Bareilly, India) 0970-4078 2229-4473 Springer Nature Singapore Singapore 536 10.1007/s42535-022-00536-7 Research Articles Assessment of the awareness about COVID-19 and the following-up of guidelines for biomedical wastes in Jaipur city Vijay Chahat chahat.vijay8306@gmail.com 1 Modi Kanak kmodi@jpr.amity.edu 2 Rajput Nitesh Singh niteshthakur72@yahoo.com 3 Sharma Vinay vsharma4@jpr.amity.edu 1 Prasad Jagdish jprasad@jpr.amity.edu 2 http://orcid.org/0000-0001-7322-584X Kulshreshtha Shweta shweta_kulshreshtha@rediffmail.com 1 1 grid.444644.2 0000 0004 1805 0217 Amity Institute of Biotechnology, Amity University Rajasthan, Jaipur, India 2 grid.444644.2 0000 0004 1805 0217 Amity School of Applied Sciences, Amity University Rajasthan, Jaipur, India 3 grid.444644.2 0000 0004 1805 0217 Amity School of Engineering and Technology, Amity University Rajasthan, Jaipur, Rajasthan India 15 12 2022 19 15 4 2022 23 11 2022 25 11 2022 © The Author(s) under exclusive licence to Society for Plant Research 2022, Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. In this COVID-19 era, isolating people and reviewing their contacts has proven to be insufficient to control the COVID-19 pandemic as there was a huge gap between exposure to the virus and isolation due to the late onset of symptoms. This led to the spread of infection and people faced the consequences not only of viral infection, but also of financial and occupational crises. People followed best management practices, however, new variants emerged that caused infection. With little information on new COVID-19 variants and their transmission, the disease spread rapidly in humans. Until now, the link between the spread of COVID-19 and the disposal of biomedical waste with household waste has not been established. Therefore, the only way to prevent infection is to make people aware. It is still necessary to open the doors for research to find the possible cause of the appearance of a new variant of COVID-19. To cope with the situation, the level of awareness among the public and their action towards the prevention of spread of infection caused by COVID-19 and its emerging variants must be known. Therefore, a survey was conducted in Jaipur from January to February 2022 to find out the status of awareness. Results of the survey revealed that both people are aware about the infection caused by COVID-19 and its variants. They are also aware about the precautions to be followed to protect themselves from acquiring COVID-19 infection. Most of the people are using masks but not gloves to prevent themselves from the infection. Merely, 71.6% of young, 100% of adults, 40% of old people sanitize their masks and gloves before disposal. Only 66.5% people are using separate bags for the collection of wastes. Despite of awareness about biomedical waste, 25% of young never sanitize, and 26.13% of young seldom sanitizes their waste before disposal. Such types of cases were not observed in adults and old age groups. Similarly, 2.3% of young did not sanitize PPE kit prior to disposal. Results of this study revealed that there is awareness about the different strains of corona virus and biomedical wastes. However, some people showed casual behaviour in the waste disposal practices. The strict implementation of rules to dispose biomedical waste will be useful for dealing with biomedical waste in this pandemic period. Keywords COVID-19 Biomedical waste Awareness Gloves Masks ==== Body pmcIntroduction Due to the disposal of surgical masks, face shields, gloves, shoe covers, and PPE, waste production has increased since the coronavirus epidemic started (PPE). According to Chandrappa and Das (2012), these wastes are known as biomedical waste (BMW) and fall into four main categories: infectious, hazardous, radioactive, and general biomedical waste. Gloves, masks, and PPE kits were used more frequently during the COVID-19 disaster and were afterwards thrown as waste (Vijay et al. 2022). However, these items could be contagious and require proper disposal and treatment. The amount of biological waste produced during the Corona virus outbreak and afterward differed significantly. In 2016, approximately 16,000 kg of biomedical waste was generated per day, which increased dramatically to 20,400 kg per day by 2020 and has now reached 23,500 kg per day in 2021 (https://timesofindia.indiatimes.com/city/hyderabad/t-stares-at-biomedical-waste-crisis-amid-covid/articleshow/84533758.cms). Because it is highly contagious, biomedical waste produced at blood banks, research labs, hospitals, clinics, and nursing homes are disposed of by incineration in accordance with the biomedical waste management guidelines. Contrarily, when biomedical waste is dumped with domestic waste by the public without being separated, it poses intriguing issues. Home-quarantined COVID-19 positive patients who were asymptomatic, disposed off their gloves and masks in dustbins, resulting in the contamination of household garbage with hazardous biomedical waste (Reddy 2020; Sangkham 2020). Improper disposal of masks and gloves by healthy and asymptomatic people after usage, which left them lying all over the street and infected nearby water bodies (Xiang et al. 2020). According to earlier surveys, 70% of participants disposed of their used masks and gloves in the trash after using them (Mejjad et al. 2021). Untreated and incorrectly disposed biomedical waste from the public can be dangerous and transmit disease across society (Healthcare waste (who.int)). In addition to increasing microbial burden, this could also impact the environment by introducing additional plastic and microplastic debris into terrestrial and aquatic ecosystems in the form of discarded masks and gloves. The issue of safety and the long-term effects on the environment are also brought up by this activity (Xiang et al. 2020). BMW is dealt by skilled staff in health care facilities, hospitals, and research facilities and disposed of at its source by incineration, chemical disinfection, wet and dry thermal treatment, microwave irradiation, land disposal, and conversion into inert material (Datta et al. 2018). However, there is no strategy in place to handle BMW emanating from residences that produce household garbage. BMW is often burned in incinerators that were operating at full capacity when COVID was in place for the disposal of BMW. Mixing garbage presented further challenges because incinerators were overloaded beyond capacity. Such situation was reported in Delhi, Vijaywada, and West Bengal, where these ran out of their capacities (Reddy 2020). In addition to the rise in disease burden, improper waste management exacerbated BMW's burden (Datta et al. 2018; Reddy 2020). The proper disposal of gloves and masks seems to be another problem. Due to the rapid increase in BMW, several nations have failed to properly handle and dispose of it, posing a risk to front-line employees engaged in handling and treatment as well as to the environment (Datta et al. 2018). In many developing countries, the proper disposal of discarded masks and gloves has been neglected, and no guidance has been given to the public (Poudel 2021). Waste can be recovered rather than disposed off based on the 3Rs idea, which stands for reduce, recycle, and reuse (ref 2). Reusability options were suggested, such as use of cloth masks repetitively after washing, employing dry heat pasteurization to disinfect N95 respirators and surgical facemasks, to reduce the amount of BMW (Kalina et al. 2022). Additionally, in order to decrease the amount of masks that have been discarded, these were hydrothermally liquefied into renewable fuel oil while ethanol was also being produced (Xiang et al. 2020). These were discarded in the pit latrine disposal system in Morocco (Kalina et al. 2022). Many nations have strict guidelines and systems in place for properly disposing of waste, such as sorting and disinfecting it first, then leaving it for nine days before sorting to reduce the risk of exposing first-line employees to viruses (Das et al. 2021). Only a small number of developed nations, though, were able to hit this goal. The way people in the community used and disposed of their gloves and masks has an influence on the production of BMW and other environmental effects (Xiang et al. 2020; Mejjad et al. 2021). The management of BMW must take into consideration in human behaviour. By raising awareness about the need for appropriate disposal of masks and gloves, individuals can inspire others to follow the same path (Poudel 2021). The current study was designed to determine the level of public awareness of the COVID-19 pandemic, biomedical waste, and methods for its disposal. This was a questionnaire-based survey carried out in February 2022. A total of 105 non-experts were selected to fill out the questionnaire about bio-medical waste management. The purpose of this study was to determine the extent of awareness of people about infection caused by COVID-19 and its variant, and biomedical waste management. Despite of their awareness, the precautions they follow to dispose off their used masks and gloves were recorded. Further, the views of the people about the amendment in waste transportation vehicle and installation of small incinerators in the society to overcome the problem of biomedical waste at societal level were listed. These suggestions of the people will help the government in improving their policies to control the release of biomedical waste at its source. Methodology Study design/questionnaire preparation The questionnaire was developed according to the other surveys conducted by many scientists (Opalinski 2008), (Hone and El Said 2016) (Krithiga et al. 2021). However, the survey included several questions to test various hypotheses that are being tested in this study, which were related to the COVID-19 pandemic and biomedical waste and its disposal practices. The study was cross-sectional and used anonymous online feedback. The questions of this survey were categorised into three categories: (i) based on their knowledge about COVID-19; (ii) based on their knowledge about biomedical wastes; and (iii) based on the disposal of biomedical waste like gloves and masks. Data collection An online survey was conducted using Google Forms in February 2022. It was circulated among the people of Jaipur with a request to fill them. Prior to administering the questionnaire, the purpose of the study was explained to all participating respondents, and their consent were taken. The filled-up responses were divided into three age groups: 14–25 years old (young people), 26–40 years old (middle-aged people), and 40–60 years old (old people). A cross-tabulation was performed to determine the relationship between their knowledge of biomedical waste and its disposal practices. Data analysis All statistical tests were performed and found to be adequate to assess the significance of differences. The sample included a random subset of non-expert individuals, and thus, the survey can be considered a representation of a larger population. There were three categories of the questionnaire and respondents were asked to fill out the form for all categories. In this questionnaire-based survey, the different choice-based questions were used to enable the participants to respond based on their agreement with a particular choice. To assess the level of awareness, all responses were clubbed in to one variable. The respondents, who have responded for all the three categories has been put as “1” otherwise “0” (Golandaj and Kallihal 2021). As the resulting dataset would be coded, collated, and analysed in SPSS 17 at a 0.05% significance level, the Chi-square test was performed. The p-values of < 0.01 and < 0.05 were considered to observe statistically significant association between a dependent variable and predictor variables. Cross-tabulation was also used to tabulate some specific questions to find out the corelation of two questions. Results and discussion Awareness to COVID-19 The results of the COVID-19 awareness related questionnaire are presented in Fig. 1. To answer the first question, "What precaution do you use frequently to avoid COVID-19 infection?", maximum of 60–80% of people in all three age groups accepted that they were using masks. Many participants (20%) preferred hand sanitizers to soap, whereas the elderly (40%) preferred soap. In response to the second question, which was related to the use of sanitizer, when COVID-19 cases are reducing, 83–100% of participants of all age groups accepted the use of sanitizers and disinfectants. About 36.4–40% of all age groups were also found to be aware of the harmful effects of sanitizer, as they mentioned in response to question 3 (Fig. 1).Fig. 1 Responses of questionnaire to analyse the awareness to COVID-19 and precautions to avoid infection among the people In response to question 4, which was related to awareness, 80% of people from all age groups were found to be aware of the fact that they could acquire infection even after vaccination. When were they asked (question 5): "What precautions do you take after being vaccinated?" 80% of the young and old age groups and 100% of adults responded that they used to wear masks despite being vaccinated. About 9.1% of youth and 20% of older people used sanitizer to clean their hands. This demonstrates the level of awareness among all age groups regarding prevention from COVID-19 infection. Everyone was aware of the precautions, such as wearing masks and using sanitizer and soap to clean and wash hands. They were aware of the spread of COVID-19 and therefore followed all precautions even after vaccination. Awareness about biomedical waste and its disposal The questionnaire used in the study contains some questions on biomedical waste and its disposal methods to find out the level of awareness of biomedical waste and its disposal. These are presented in Fig. 2. "Are you aware of biomedical waste?" asked question 6; 89.8% of young people, 90% of adults, and 60% of the elderly were aware of biomedical waste. Every day, 18.2% of participants discard masks. However, 39.8% of participants used cloth masks, which were washed and reused (Question 7, Fig. 2). However, only 71.6% of young people, 100% of adults, and 40% of the elderly sanitise their masks and gloves before throwing them away (question 8, Fig. 2). It is pertinent to note that 31.8% of the young participants never sanitized their mask before disposal (Question 9, Fig. 2).Fig. 2 Questions asked to the participants related to biomedical waste i.e., masks/gloves disposal A cross-tabulation (Fig. 3) of people's awareness of biomedical waste (Question 6, Fig. 2) and the frequency of washing or sanitising masks before disposal (Question 9, Fig. 2) revealed that 6.81% of the young are unaware and, as a result, do not sanitise their masks before disposal (Fig. 3). Despite their awareness of biomedical waste, 25% of young people never sanitised their waste before disposal, and 26.13% only rarely sanitised it. Such types of cases were not observed in adults or older age groups (Question 9, Fig. 2). Similarly, 2.3% of the young did not sanitise PPE kits prior to disposal (Question 10, Fig. 2). This shows the casual behaviour of young people towards the disposal of gloves and masks, despite their awareness about biomedical waste.Fig. 3 Cross-tabulation between the questions to show the relationship of participant’s awareness and their action to wash or sanitize their masks (* reveals significant result i.e. p < 0.05) Despite of their awareness on biomedical waste, 31.8% of young people, 40% of adults, and 60% of the elderly did not use separate bags for waste disposal (Question 11; Fig. 4). To determine their level of awareness regarding the use of separate waste bags, 5.68% of adults and 20% of the elderly were found to be unaware and, as a result, did not use separate waste bags (p 0.05) (Fig. 4). In order to avoid the risks posed by any person infected with COVID-19, they were kept in isolation and provided with appropriate medical care (Question 12, Fig. 5). The result of the cross-tabulation of question 6 with question 11 revealed that participants didn’t used separate bags for waste disposal (Question 11, Fig. 6), despite their awareness. This is a matter of concern, as improper disposal may lead to the spread of infection through infected materials and may give rise to mutants. Only 2% of young people used personal protective equipment (PPE) kits and disposed of them in separate bags (p < 0.05) (Fig. 6). This depicted the lack of awareness about the use of separate bags for disposing of infected material, i.e., biomedical waste. These findings are in contrast to the study conducted in Tamil Nadu about biomedical waste, which reported the use of separate containers by 72.8% of health care workers (Krithiga et al. 2021) (Dalui et al. 2021).Fig. 4 Questions asked to the participants related to biomedical waste and their way of avoidance to risk Fig. 5 Cross-tabulation to know the relationship between the participants’ awareness about the medical waste and use of separate bags for different types of wastes (* reveals significant result i.e. p < 0.05) Fig. 6 Cross-tabulation to know the relationship between the participants’ awareness about the way of avaoiding the risk of COVID-19 and use of separate bags for the disposal of waste (* reveals significant result, p < 0.05) People were aware of biodegradable masks, and thus, they would purchase them regardless of their price (Question 14, Fig. 4). This reveals the need of training the public about biomedical waste and the disposal of masks after use. Suggestions to Government The people were asked about their views on including a separate section in a waste transportation vehicle for biomedical waste, 87.5% of youth, 100% of adults, and 100% of old age persons agreed, and 5.7% of youth remained unanswered (Question 14, Fig. 7). When asked, "Should the government include a section of a medical waste transportation vehicle?" 100% of adults and seniors agreed that a separate section in a waste transportation vehicle should be included. However, 90.99% of people agreed that the government should incorporate a separate section in waste transportation vehicles (Question 15, Fig. 7).Fig. 7 Questions related to their suggestion to control the exposure of waste to the workers dealing with it A cross-tabulation of these two questions revealed that 100% of adults and older people desired a separate section for biomedical waste disposal in waste transportation vehicles. Only 2.27% people were unaware of this and not recommended it to the government (p < 0.05) (Fig. 8). This data showed that people were aware of the segregation of waste in waste transportation vehicles. There has been no report published on the public's demand for a separate section for biomedical waste, as well as the need for training and awareness in this area.Fig. 8 Cross-tabulation of two questions i.e., should a separate section or container be included in waste transportation vehicle with should government include a separate section in waste transportation vehicle In addition to this, 90% of youth, and 100% of adults and older people preferred the installation of incinerators in colonies for masks and gloves to prevent the spread of infection (Question 16, Fig. 7). This reveals the status of disposing masks and gloves among them by incineration. Conclusion Biomedical waste, generated by the public during the COVID-19 pandemic, is an important problem that needs to be addressed properly, and the public must be aware of it. Improper disposal and ignorance of the waste accumulation may give rise to several problems, including the spread of infection. Awareness among the people can be a major tool to combat issues related to the disposal of masks and gloves during the COVID-19 pandemic. Regardless of the person's symptoms, waste disposal requirements must be properly observed. This study reveals that people are aware of the pandemic and biomedical wastes, their consequences, and the precautions that need to be taken to cope with the situation. A broad public awareness campaign may be beneficial in making people aware of the importance of strictly adhering to biomedical waste guidelines. Acknowledgements We are thankful to the Director, Amity Institute of Biotechnology, Amity University Rajasthan, Jaipur for providing all the facilities for conducting the present work. We are also thankful to the Central Instrumentation facility of University of Rajasthan, Jaipur for providing FTIR facility. Funding This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors. There is no declaration related to this manuscript. Data availability The datasets generated during and/or analysed during the current study are available from the corresponding author on reasonable request. Declarations Conflict of interest There are no competing interest related to this manuscript. Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References Chandrappa R, Das DB (2012) Biomedical waste. ln: Solid Waste Management. Environ Sci Eng 9783642286803:147. 10.1007/978-3-642-28681-0_6 Dalui A Banerjee S Roy R Assessment of knowledge, attitude, and practice about biomedical waste management among healthcare workers during COVID-19 pandemic in a health district of West Bengal Indian J Public Health 2021 65 4 345 10.4103/IJPH.IJPH_2103_21 34975076 Das AK Islam MN Billah MM Sarker A COVID-19 and municipal solid waste (MSW) management: a review Environ Sci Pollut Res Int 2021 28 23 28993 29008 10.1007/s11356-021-13914-6 33877522 Datta P Mohi GK Chander J Biomedical waste management in India: Critical appraisal J Lab Phys 2018 10 1 6 14 10.4103/JLP.JLP_89_17 Golandaj JA Kallihal KG Awareness, attitude and practises of biomedical waste management amongst public health-care staff in Karnataka, India Int J Humanit Appl Soc Sci 2021 3 1 49 63 10.1108/JHASS-08-2019-0041 Health-care waste (who.int) Accessed on 2nd April 2022 Hone KS El Said GR Exploring the factors affecting MOOC retention: a survey study Comput Educ 2016 98 157 168 10.1016/J.COMPEDU.2016.03.016 Kalina M Kwangulero J Ali F Tilley E “You need to dispose of them somewhere safe”: Covid-19, masks, and the pit latrine in Malawi and South Africa PLoS ONE 2022 17 2 e0262741 10.1371/journal.pone.0262741 35192618 Krithiga P, Sudharsana V, Sribalaji R, Snega C. Covid 19 (2021) Pandemic: Assessment of Knowledge and Attitudes in Biomedical Waste Management among Health Care Professionals in Tamil Nadu. Asia Pacific J Heal Manag. 16(3):154–164. 10.24083/APJHM.V16I3.987 Mejjad N Cherif EK Rodero A Krawczyk DA El Kharraz J Moumen A Laqbaqbi M Fekri A Disposal behavior of used masks during the COVID-19 pandemic in the Moroccan community: Potential environmental impact Int J Environ Res Public Health 2021 18 8 4382 10.3390/ijerph18084382 33924217 Opalinski L Older Adults and the Digital Divide: Assessing Results of a Web-Based Survey J Technol Hum Serv 2008 18 3–4 203 221 10.1300/J017v18n03_13 Reddy A (2020) Biomedical Waste in India has Increased Exponentially as a Result of COVID-19 (vidhilegalpolicy.in) accessed on 12th November 2022 Sangkham S Face mask and medical waste disposal during the novel COVID-19 pandemic in Asia Case Studies in Chemical and Environmental Engineering 2020 2 100052 10.1016/j.cscee.2020.100052 Sonika P (2021) Disposing of face masks: The next environmental problem? | UNICEF Nepal. accessed on 12th November 2022 Sribala V (2021) Telangana stares at biomedical waste crisis amid Covid. https://timesofindia.indiatimes.com/city/hyderabad/t-stares-at-biomedical-waste-crisis-amid-covid/articleshow/84533758.cms. Accessed 01 April 2022 The Third Pole (2021) Biomedical waste and Covid-19: 'People on the front line are facing serious issues' (thethirdpole.net) accessed on 12th November 2022 Vijay C, Sharma A, Kulshreshtha S (2022) Heeded and Unheeded Attentions about Covid-19 Viral Transmission with Special Reference to Biomedical Wastes. Int j biol pharm allied sci 11(1):48–58. 10.31032/IJBPAS/2022/11.1.5805 Xiang Y Song Q Gu W Decontamination of surgical face masks and N95 respirators by dry heat pasteurization for one hour at 70 °C Am J Infect Control 2020 48 880 882 10.1016/j.ajic.2020.05.026 32479844

0

PMC9753887

NO-CC CODE

2022-12-16 23:26:21

no

Vegetos. 2022 Dec 15;:1-9

utf-8

Vegetos

10.1007/s42535-022-00536-7

oa_other

==== Front Reprod Sci Reprod Sci Reproductive Sciences 1933-7191 1933-7205 Springer International Publishing Cham 36520404 1082 10.1007/s43032-022-01082-y Reproductive Biology: Original Article Hyperbaric Oxygen Treatment Ameliorates the Decline in Oocyte Quality and Improves the Fertility of Aged Female Mice Ma Yang 1 Zhong Yanyu 2 Chen Xia 1 Liu Huijun 1 Shi Yichao 1 Zhang Xiuwen 1 http://orcid.org/0000-0001-5018-5769 Sun Huiting 94sunhuiting@163.com 1 1 grid.89957.3a 0000 0000 9255 8984 Department of Reproductive Center, The Affiliated Changzhou No. 2 People’s Hospital of Nanjing Medical University, No. 68 Gehu Road, Changzhou, 213003 Jiangsu China 2 grid.429222.d 0000 0004 1798 0228 Department of Reproductive Center, The First Affiliated Hospital of Soochow University, Suzhou, Jiangsu China 15 12 2022 17 24 5 2022 8 9 2022 © The Author(s), under exclusive licence to Society for Reproductive Investigation 2022, Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. The age-related decay in oocyte quality contributes to the gradual decline in fertility and the final occurrence of natural sterility. In this study, we aimed to investigate the effects of the hyperbaric oxygen treatment (HBOT) on oocyte quality in aging mouse oocyte. Eight- and forty-week-old female C57BL/6 J mice were treated with HBO for 10 days, and the quality of oocytes was analyzed. The results revealed that HBOT improved the age-related serum AMH levels. While compared with untreated aged mice, HBOT showed reduced follicular apoptosis and improved oocyte maturation, fertilization, and blastocyst formation in aged mice. HBO triggered changes in the microRNA expression in the ovaries of aged mice. In this study, 27 DEGs were identified in the HBOT mouse ovarian tissues, of which 9 were upregulated and 18 were downregulated. Notably, KEGG analysis revealed that these genes involved in different biological processes differed significantly in the ovary. Among these, the PI3K-Akt signaling was the most prominent pathway that controlled the recruitment and growth of primordial follicles. The calcium signaling pathway was found to be involved during the peri-implantation period. These results suggest that HBOT can be applied to improve the quality of oocytes, and it could be a potential clinical application to improve the fertility of aged female. Keywords Hyperbaric oxygen Aged Female Mice Ovarian reserve MicroRNA The present study was supported by the Changezhou Sci & Tech ProgramCJ20200103 CJ20220077 Sun Huiting ==== Body pmcIntroduction It has been well known that with increasing chronological age, female fecundity decreases. In view of the current trend to postpone childbearing in contemporary populations, the age-related decrease of the ability to produce offspring in female has gradually appeared [1]. The mechanisms underlying the observed gradual decline of the follicle pool and the reduced oocyte quality are far from being fully understood. Decreasing numbers of follicles, coinciding with diminished oocyte quality, dictates the gradual changes in menstrual cycle regularity and monthly fecundity [2–4]. Recent knowledge about the age-related female fecundity declines was mainly because the risk of oocyte chromosome abnormalities and with increases adverse pregnancy outcomes [5, 6]. Hyperbaric oxygen treatment (HBOT) is the application of 100% oxygen at environmental pressures > 1 atmosphere [7, 8]. HBOT, generally used as an effective treatment method in ischemia reperfusion injury, anti-inflammatory effects and stimulates the antioxidant system. It has been widely applied to accelerate fracture healing [9], articular cartilage injury [10], type 2 diabetes mellitus [11, 12], and the proliferation of epidermal basal cells [13]. Accumulation of oxides in the ovarian tissues in older women may be one of the reasons for infertility, considering that it reduces the quality of embryos. In male infertility, sperm DNA fragmentation was improved after HBOT [14], and for male recover, erectile function has also been ameliorated [15, 16]. On another hand, HBOT can significantly improve female endometrial receptivity in the cycle for better outcome of pregnancy implantation [17]. Additionally, some research inferred that exposure to mild HBO can increase the oxygen supply to the tissues, thereby enhancing oxidative metabolism [11]. Overall, HBOT provides the therapeutical efficacy in many diseases, but it in female ovary influence has rarely known. In this work, we investigated the effect of HBOT on the ovarian reserve of 40-week-old mice. Moreover, miRNA profiling revealed a possible compensatory mechanism of HBOT in age-related decline in ovarian functions. Our results provided a theoretical basis for the clinical treatment of age-related decline with ovarian function. Materials and Methods Mice and HBO Therapy Female C57BL/6 J mice aged 8 and 40 weeks were sourced from the Carvens Animal Laboratory Technology Company [http://www.cavens.com.cn/index.php; animal certificate no. SCXK(SU)2016–0010]. All animal-use experiments were approved by the Animal Care and Use Committee of Nanjing Medical University. The experimental mice were kept in a standard pathogen-free environment under standard housing conditions. The room was maintained under a strictly controlled temperature (22 ± 1 °C) and humidity (52 ± 5%). The mice were provided ad libitum feed with a 12-h light/dark photophase condition. The mice were weighed weekly and their health status was monitored. For the HBOT, the mice were administered 100% oxygen at a pressure of 2.5 ATA in a custom-made mono-chamber intended for small animals for 90 min daily for 10 consecutive days. PMSG Injection for Super-ovulated Mice Female C57BL/6 mice (OC and OC + HBOT groups) were intraperitoneally injected with 5 IU PMSG (Prospec Tany, TechnoGene, Ltd.) to induce super-ovulation. Next, for in vitro fertilization (IVF) and embryo culture, the mice were sacrificed through cervical dislocation 48 h later after the PMSG injection. The IVF and embryo culture were performed as described elsewhere [18]. Immunofluorescence Staining The ovarian slides were stained with Alexa Fluor 488 (1:200; Beyotime Biotechnology, China) overnight under 4 °C. The next day, the slides were incubated at room temperature in the dark with Alexa Fluor 594-conjugated goat anti-mouse IgG (Life Technology, Waltham, MA, USA). Hoechst was used to stain the nuclei. Images were collected through fluorescent microscopy (Olympus, U-RFL-T). RNA Extraction and Quantitative Real-Time PCR (qPCR) Total RNA was separated and extracted by the RNeasy Mini Kit according to the manufacturer’s protocol (Vazyme Biotech, Nanjing, China). Then, 1.0 μg RNA was used as a template for cDNA synthesis by using the HiScript III 1st Strand cDNA Synthesis Kit (+ gDNA wiper) (Vazyme Biotech). The cDNA samples were analyzed by qPCR using the AceQ qPCR SYBR Green Master Mix (Low ROX Premixed). U6 snRNA was used as an internal control. The experimental data were analyzed by the relative quantification method (2-∆∆Cq) [19]. All experiments were repeated thrice. Statistical Analyses Original data analysis was performed with GraphPad Prism 5.01 (GraphPad Software, Inc.) and ImageJ (National Institutes of Health). Data are presented as the means ± SD (n = 3). The differences between the two groups were analyzed by unpaired Student’s t-test. P < 0.05 was considered to indicate statistical significance. Result HBOT improved age-related decline of the serum AMH and FSH levels. The bodyweight of the mice in each group was evaluated after 10 cycles of HBOT (Fig. lA). HBOT showed no significant change in bodyweight (Fig. lB). However, HBOT mice showed significantly improved serum FSH levels in the OC + HBOT group (Fig. 1C), although the serum LH levels did not change. OC mice (40-week-old) showed lower serum AMH levels, possibly due to decreased ovarian reserve (Fig. 1D). HBOT mice showed significant improvement in their serum AMH levels.Fig. 1 Bodyweight variation and serum hormone levels of mice after HBOT. A 40-week-old mice in the aging control group (n = 6, OC group), and 40-week-old mice were treated with HBO 9 times each day (n = 6, OC + HBOT group). Post-HBOT, the changes in the mice bodyweight were measured along with hormonal assessment. B Bodyweight variations. C Follicle-stimulating hormone (FSH) and luteinizing hormone (LH) serum levels. D 8-week-old mice as young controls (n = 6, YC group). Anti-Mullerian hormone (AMH) serum levels. Data are presented as mean ±standard error of the mean. Statistical analysis was performed using Student’s t-test Treatment with HBO ameliorates the quality of aged oocytes in vivo In order to explore the effects of HBOT on age-related decline in oocytes in vivo, 3 groups of mice were treated with super-ovulation via hormonal stimulation, and IVF experiments were conducted using super-ovulated oocytes obtained from 3 groups of mice (Fig. 2A). Hence, we assessed the quantity and quality of oocytes. In addition to assessing the morphology of oocytes, the number of oocytes retrieved from the OC group was also significantly lower than that of the YC group, while that of the OC + HBOT group was closer to that of the YC group in terms of the number of oocytes retrieved (P < 0.05; Fig. 2C). The number of fertilizations performed was significantly higher in the OC + HBOT mice than in the OC mice (Fig. 2C). Taken together, these results indicate that exposure to HBO ameliorates the oocyte quality and can hence enhance the quality of fertilization in age-related mice.Fig. 2 Quality and number of mouse oocytes retrieved after HBOT. A Eight-week-old mice as young control (n = 6, YC group); 40-week-old mice as aging control (n =6, OC group); and 40-week-old mice treated with HBO 9 times each day (n = 6, OC+HBOT group). After HBOT, the mice were induced with hormonal stimulation to implement the IVF experiments. B Typical images of MII oocytes were collected from YC (n = 20), OC (n = 11), and OC + HBOT (n = 22) mice, and the IVF outcomes from these 3 groups. Scale bar = 100 μm, 400 μm. C Oocytes were retrieved from the YC, OC, and OC + HBOT groups after 18 h of PMSG injection. The number of retrieved oocytes and 2-cell embryo (2PN), fertilization, cleavages, and blastocyst formation. Data are presented as the mean ± standard error of the mean. Statistical analysis was performed using Student’s t-test HBOT changes in microRNA expression in aged mouse ovaries To further explore the potential mechanisms underlying the influence of HBOT, deferentially expressed miRNA profiles in the OC and OC + HBOT groups were compared using RNA sequencing technology. A deep interpretation of the miRNA networks can help elucidate the mechanisms of improvement in the quality of aged oocytes and facilitate the development of new treatment strategies for elderly patients. A total of 27 DEGs were identified in HBOT mice testes by applying the cutoffs of FC > 2 and FDR ≤ 0.05, 9 of which were upregulated and 18 were downregulated (Fig. 3A). Notably, KEGG analyses revealed that these genes involved in different biological processes were remarkably divergent in the ovary (Fig. 3B).Fig. 3 Volcano plot analysis of differentially expressed miRNAs and the pathway enrichment scatter plot of differential microRNA target genes. A RNA sequencing analysis was performed to compare the gene expressions between the OC (n = 3) and OC + HBOT (n = 3) groups. We found that 9 DEGs were upregulated and 18 were significantly downregulated. B The pathway enrichment scatter plot of differential microRNA target genes. The pathways involved in the differential microRNA target genes. The size of the dots indicates the number of differential microRNA target genes in this pathway, and the colors of the dots indicate different Q values. The greater the rich factor with the better relatives HBOT reduces follicle apoptosis The expressions of two DEGs were determined by qPCR. MiR-99b-5p, miR3103-3p, miR3074-1-3p, and miR211-5p were found to have a certain influence in HBOT mice (Fig. 4A). To confirm this effect, DNA fragmentation in the HBOT ovary was stained to assess the quality of aged follicles (Fig. 4B). We have presented a summarized list of known research on miRNAs involved in aged mice after HBOT in Table 1, which includes the combinations of miRNAs and their confirmed target genes. This list is critical to the present research topic.Fig. 4 qRT-PCR validated differentially expressed micro-RNAs and the number of aged follicles after HBOT. A The expression of micro-RNA was measured in the OC and OC + HBOT groups. B Follicle quality was assessed by examining the stained nuclei and DNA fragmentation in each mice group. C The quantity of aged follicles was assessed from apoptotic nuclei. Data are presented as mean ± standard error of the mean. Statistical analysis was performed using Student’s t-test Table 1 miRNAS target genes and thier functions in HBO treatment mice micro_ID GeneSymbol mmu-miR-99b-5p Dip2a; Ppp2r5a; Gatm; Arhgef1; Ablim1; Polrmt; P2rx5 mmu-miR-3103-3p Zdbf2; Lrrc3; Mrtfa; Rps6ka3; Nfam1; ltga6; Kcnmb4; Spry1; Zfp280b; Erp27; Picxd2; Dedd2; Spsb2; Crem; Lemd1; Zfp408; Dlc1; Map4k4; Hmx2; Orc1; Myoz3; Ndst1; Rbm4b; Parqr4; Adora1; Tctn2; Gcn1; Slc4a4; Klra13-ps; Gm18761; Gm18337 mmu-miR-3074-1-3p Cers3; Aspa; Angpt4; Cntn6; Rbfa; Med28; S1pr2; Aopep mmu-miR-211-5p Gm9747; Pip4k2b; Ccdc50; lvl; Zfp652; Tanc1; Socs6; Tnpo2; Slc9a2; Fzd5; Nfz1; Gm17018; Gm9449 Discussion Currently, delayed childbearing brings forth a common problem in women. Ovarian aging in women correlates with a decrease in the number of ovarian follicles and declining oocyte quality. As a result, reproductive aging in females brings the significant increase in the risk of adverse reproductive outcomes, like higher rates of spontaneous abortion and miscarriage and a significant decline in oocyte quality [19–23]. With most recently research, dehydroepiandrosterone and growth hormone may slightly improve the clinical pregnancy rate, albeit the data for supporting these therapies are insufficient [24–26]. In this study, we investigated the effects of the hyperbaric oxygen therapy on oocyte quality in aging mouse and found that HBOT could improve the quality of oocytes. Hyperbaric oxygen therapy (HBOT), a therapeutical method based on exposure to pure concentrations of oxygen (O2), can increasing the concentration of oxygen in the blood and tissues. Thus, HBOT provides multiple effects in the organism, including in the treatment of diabetic foot ulcers [27, 28], sports musculoskeletal injuries [29], novel coronavirus infection [30], and so on. HBOT has been indicated positive value by evidence-based medicine distinctly. The mouse is widely used as a model for human biological aging [31–33]. It is commonly agreed that the absolute age of reproductive biological aging in mice between 9 and 15 months [34, 35]. Accordingly, our research used 40-week-old female C57BL/6 J mice as aging model and revealed that the oocyte quality and fertilization can be improved through exposure to HBO. Moreover, follicle apoptosis was reduced in the HBOT group probably due to the PI3K/Akt or JAK/STAT signaling pathway according to the KEGG enrichment analysis. These data suggest that there is multiple influence for aging follicle development by HBOT. AMH, which has a stable level in serum, is secreted by granulosa cells in late preantral follicles and small antral follicles, and it is more closely related with the number of primordial follicles [36]. Hence, AMH offers a higher predictive value than FSH and E2 level in clinical ovarian reserve decline patients [37, 38]. More than that, FSH is directly regulate on follicle development [39]. Interestingly, our study observed that HBOT improved the serum AMH and FSH level. And, the slightly increase FSH level may directly stimulate ovarian granulosa cells to promote follicle growth and further development. Current research has been concentrated upon determining the factors that are most important to ovarian health. One goal of these studies is to elucidate the molecular mechanisms underlying the changes in ovarian reserve, oocyte viability, and oocyte quality [40]. Insulin-like growth factor (IGF), which regulates the stimulatory effect of FSH on aromatase expression, is the most important local factors in the system [41]. In addition, insulin-like growth factor-1 receptor (IGF1R) is essential for follicle survival and fertility in female mice ovary [42]. In our study, we identified several miRNAs changes by RNA-seq, as expected, miR-99b-5p, an upstream target for IGF1R [43]. In aging oocytes, the raised DNA damage is closely related to apoptosis, poor quality of oocytes, and eventually caused infertility and miscarriage. In the presence of DNA damage, cells activate a coordinated mechanism called the DNA damage response (DDR) to activate different repair processes to correct the damage [40]. There is evidence to support the existence of crosstalk between the PI3K/Akt signaling pathway and the DDR in cells [42, 44]. High PI3K/Akt activity is linked up with a decline in the number of primordial follicles and ovarian aging [45]. Ovarian aging is associated with impaired DDR within oocytes [46]. In our work, KEGG analysis showed that these genes were involved in the PI3K-Akt signaling pathway biological processes. DNA fragmentation stain in the HBOT ovary was showed effectively ameliorated by immunofluorescence. These data together suggest that multiple genes of the primordial follicle recruitment pathway are influenced by HBOT. In conclusion, HBOT may contribute to the improvement of oocyte quantity in the aged mouse model, although further research and clinical trial are needed to explore and potential therapy for aging female fecundity decreases. Funding The present study was supported by the Changezhou Sci &Tech Program (grant no. CJ20220077 and CJ20200103). Data Availability The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request. Code Availability Any software application or custom code described in the manuscript is available for testing by reviewers in a way that preserves their anonymity. Declarations Ethics Approval All animal-use experiments were approved by the Animal Care and Use Committee of Nanjing Medical University. Consent to Participate All experiments were conducted according to the National Institute of Health guidelines on the care and use of animals and approved by the Animal Care and Use Committee of Nanjing Medical University. Consent for Publication For the manuscripts that include details, images, or videos relating to animals, written informed consent for the publication of these details was approved by the Animal Care and Use Committee of Nanjing Medical University. Conflict of Interest The authors declare no competing interests. Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References 1. Trawick E Pecoriello J Quinn G Goldman KN Guidelines informing counseling on female age-related fertility decline: a systematic review J Assist Reprod Genet 2021 38 1 41 53 10.1007/s10815-020-01967-4 33188440 2. Broekmans FJ Soules MR Fauser BC Ovarian aging: mechanisms and clinical consequences Endocr Rev 2009 30 5 465 493 10.1210/er.2009-0006 19589949 3. Cimadomo D Fabozzi G Vaiarelli A Ubaldi N Ubaldi FM Rienzi L Impact of maternal age on oocyte and embryo competence Front Endocrinol (Lausanne) 2018 9 327 10.3389/fendo.2018.00327 30008696 4. Palomba S Daolio J Romeo S Battaglia FA Marci R La Sala GB Lifestyle and fertility: the influence of stress and quality of life on female fertility Reprod Biol Endocrinol 2018 16 1 113 10.1186/s12958-018-0434-y 30501641 5. Kim J You S Effect of samul-tang on female fertility via RAS signaling pathway in ovaries of aged mice Aging (Albany NY) 2021 13 11 14829 14842 10.18632/aging.203150 34091440 6. Busnelli A, Navarra A, Levi-Setti PE. Qualitative and quantitative ovarian and peripheral blood mitochondrial DNA (mtDNA) alterations: mechanisms and implications for female fertility. Antioxidants (Basel). 2021. 10(1). 7. Chiu CH Chang SS Chang GJ The effect of hyperbaric oxygen treatment on myoblasts and muscles after contusion injury J Orthop Res 2020 38 2 329 335 10.1002/jor.24478 31531986 8. Bennett M, Best TM, Babul S, Taunton J, Lepawsky M. Hyperbaric oxygen therapy for delayed onset muscle soreness and closed soft tissue injury. Cochrane Database Syst Rev. 2005. 2005(4): CD004713. 9. Kawada S Wada E Matsuda R Ishii N Hyperbaric hyperoxia accelerates fracture healing in mice PLoS ONE 2013 8 8 e72603 10.1371/journal.pone.0072603 23967323 10. Ueng SW Yuan LJ Lin SS Hyperbaric oxygen treatment prevents nitric oxide-induced apoptosis in articular cartilage injury via enhancement of the expression of heat shock protein 70 J Orthop Res 2013 31 3 376 384 10.1002/jor.22235 22991091 11. Ishihara A Mild hyperbaric oxygen: mechanisms and effects J Physiol Sci 2019 69 4 573 580 10.1007/s12576-019-00678-5 31062232 12. Nagatomo F Takemura A Roy RR Fujino H Kondo H Ishihara A Mild hyperbaric oxygen inhibits the growth-related decline in skeletal muscle oxidative capacity and prevents hyperglycemia in rats with type 2 diabetes mellitus J Diabetes 2018 10 9 753 763 10.1111/1753-0407.12666 29633563 13. Nishizaka T Nomura T Higuchi K Takemura A Ishihara A Mild hyperbaric oxygen activates the proliferation of epidermal basal cells in aged mice J Dermatol 2018 45 9 1141 1144 10.1111/1346-8138.14484 29855082 14. ÖzgökKangal K Özgök Y Assisted reproductive treatments with hyperbaric oxygen therapy in male infertility Turk J Urol 2021 47 2 98 105 10.5152/tud.2020.20328 33201797 15. Hadanny A Lang E Copel L Hyperbaric oxygen can induce angiogenesis and recover erectile function Int J Impot Res 2018 30 6 292 299 10.1038/s41443-018-0023-9 29773856 16. Sahin MO Sen V Eser E The effect of hyperbaric oxygen therapy on erectile functions: a prospective clinical study Urol Int 2018 101 2 206 211 10.1159/000490356 30089290 17. Mitrović A Nikolić B Dragojević S Brkić P Ljubić A Jovanović T Hyperbaric oxygenation as a possible therapy of choice for infertility treatment Bosn J Basic Med Sci 2006 6 2 21 24 10.17305/bjbms.2006.3168 16879108 18. Liu C Growth hormone ameliorates the age-associated depletion of ovarian reserve and decline of oocyte quality via inhibiting the activation of Fos and Jun signaling Aging (Albany NY) 2021 13 5 6765 6781 10.18632/aging.202534 33621201 19. Livak KJ Schmittgen TD Analysis of relative gene expression data using real-time quantitative PCR and the 2(-Delta Delta C(T)) Method Methods 2001 25 4 402 408 10.1006/meth.2001.1262 11846609 20. Flenady V Koopmans L Middleton P Major risk factors for stillbirth in high-income countries: a systematic review and meta-analysis Lancet 2011 377 9774 1331 1340 10.1016/S0140-6736(10)62233-7 21496916 21. American College of Obstetricians and Gynecologists Committee on Gynecologic Practice and Practice Committee. Female age-related fertility decline. Committee Opinion No. 589. Fertil Steril. 2014;101:633–34. 22. Qin JC Fan L Qin AP The effect of dehydroepiandrosterone (DHEA) supplementation on women with diminished ovarian reserve (DOR) in IVF cycle: Evidence from a meta-analysis J Gynecol Obstet Hum Reprod 2017 46 1 1 7 10.1016/j.jgyn.2016.01.002 28403950 23. Schwarze JE Canales J Crosby J Ortega-Hrepich C Villa S Pommer R DHEA use to improve likelihood of IVF/ICSI success in patients with diminished ovarian reserve: a systematic review and meta-analysis JBRA Assist Reprod 2018 22 4 369 374 30125071 24. Cozzolino M Cecchino GN Troiano G Romanelli C Growth hormone cotreatment for poor responders undergoing in vitro fertilization cycles: a systematic review and meta-analysis Fertil Steril 2020 114 1 97 109 10.1016/j.fertnstert.2020.03.007 32553470 25. Xu YM Hao GM Gao BL Application of growth hormone in in vitro fertilization Front Endocrinol (Lausanne) 2019 10 502 10.3389/fendo.2019.00502 31396161 26. Li XL Wang L Lv F The influence of different growth hormone addition protocols to poor ovarian responders on clinical outcomes in controlled ovary stimulation cycles: a systematic review and meta-analysis Medicine (Baltimore) 2017 96 12 e6443 10.1097/MD.0000000000006443 28328856 27. Zhang Z Zhang W Xu Y Liu D Efficacy of hyperbaric oxygen therapy for diabetic foot ulcers: an updated systematic review and meta-analysis Asian J Surg 2022 45 1 68 78 10.1016/j.asjsur.2021.07.047 34376365 28. Kumar A Shukla U Prabhakar T Srivastava D Hyperbaric oxygen therapy as an adjuvant to standard therapy in the treatment of diabetic foot ulcers J Anaesthesiol Clin Pharmacol 2020 36 2 213 218 10.4103/0970-9185.74924 33013037 29. Moghadam N Hieda M Ramey L Levine BD Guilliod R Hyperbaric oxygen therapy in sports musculoskeletal injuries Med Sci Sports Exerc 2020 52 6 1420 1426 10.1249/MSS.0000000000002257 31876671 30. Thibodeaux K Speyrer M Raza A Yaakov R Serena TE Hyperbaric oxygen therapy in preventing mechanical ventilation in COVID-19 patients: a retrospective case series J Wound Care 2020 29 Sup5a S4 S8 10.12968/jowc.2020.29.Sup5a.S4 31. Briley SM Jasti S McCracken JM Reproductive age-associated fibrosis in the stroma of the mammalian ovary Reproduction 2016 152 3 245 260 10.1530/REP-16-0129 27491879 32. Chiang T Duncan FE Schindler K Schultz RM Lampson MA Evidence that weakened centromere cohesion is a leading cause of age-related aneuploidy in oocytes Curr Biol 2010 20 17 1522 1528 10.1016/j.cub.2010.06.069 20817534 33. Duncan FE Jasti S Paulson A Kelsh JM Fegley B Gerton JL Age-associated dysregulation of protein metabolism in the mammalian oocyte Aging Cell 2017 16 6 1381 1393 10.1111/acel.12676 28994181 34. Cui LB Zhou XY Zhao ZJ Li Q Huang XY Sun FZ The Kunming mouse: as a model for age-related decline in female fertility in human Zygote 2013 21 4 367 376 10.1017/S0967199412000123 23517725 35. Dutta S Sengupta P Men and mice: relating their ages Life Sci 2016 152 244 248 10.1016/j.lfs.2015.10.025 26596563 36. Josso N Women in reproductive science: anti-Müllerian hormone: a look back and ahead Reproduction 2019 158 6 F81 F89 10.1530/REP-18-0602 30844753 37. Broer SL Eijkemans MJ Scheffer GJ Anti-Mullerian hormone predicts menopause: a long-term follow-up study in normoovulatory women J Clin Endocrinol Metab 2011 96 8 2532 2539 10.1210/jc.2010-2776 21613357 38. Van Rooij IA Broekmans FJ Scheffer GJ Serum anti-Mullerian hormone levels best reflect the reproductive decline with age in normal women with proven fertility: a longitudinal study Fertil Steril 2005 83 4 979 987 10.1016/j.fertnstert.2004.11.029 15820810 39. Wallach EE Taymor ML The regulation of follicle growth: some clinical implications in reproductive endocrinology Fertil Steril 1996 65 2 235 247 10.1016/S0015-0282(16)58077-7 8566240 40. Park SU Walsh L Berkowitz KM Mechanisms of ovarian aging Reproduction 2021 162 2 R19 R33 10.1530/REP-21-0022 33999842 41. Baumgarten SC Armouti M Ko C Stocco C IGF1R expression in ovarian granulosa cells is essential for steroidogenesis, follicle survival, and fertility in female mice Endocrinology 2017 158 7 2309 2318 10.1210/en.2017-00146 28407051 42. Karimian A Mir SM Parsian H Crosstalk between phosphoinositide 3-kinase/Akt signaling pathway with DNA damage response and oxidative stress in cancer J Cell Biochem 2019 120 6 10248 10272 10.1002/jcb.28309 30592328 43. Jiang S Human bone marrow mesenchymal stem cells-derived exosomes attenuated prostate cancer progression via the miR-99b-5p/IGF1R axis Bioengineered 2022 13 2 2004 2016 10.1080/21655979.2021.2009416 35030978 44. Plo I Laulier C Gauthier L Lebrun F Calvo F Lopez BS AKT1 inhibits homologous recombination by inducing cytoplasmic retention of BRCA1 and RAD51 Cancer Res 2008 68 22 9404 9412 10.1158/0008-5472.CAN-08-0861 19010915 45. Reddy P Adhikari D Zheng W PDK1 signaling in oocytes controls reproductive aging and lifespan by manipulating the survival of primordial follicles Hum Mol Genet 2009 18 15 2813 2824 10.1093/hmg/ddp217 19423553 46. Oktay K Turan V Titus S Stobezki R Liu L BRCA mutations, DNA repair deficiency, and ovarian aging Biol Reprod 2015 93 3 67 10.1095/biolreprod.115.132290 26224004

36520404

PMC9753892

NO-CC CODE

2022-12-16 23:26:35

no

Reprod Sci. 2022 Dec 15;:1-7

utf-8

Reprod Sci

10.1007/s43032-022-01082-y

oa_other

==== Front Soft comput Soft comput Soft Computing 1432-7643 1433-7479 Springer Berlin Heidelberg Berlin/Heidelberg 7729 10.1007/s00500-022-07729-x Data Analytics and Machine Learning A novel proposed CNN–SVM architecture for ECG scalograms classification http://orcid.org/0000-0001-9841-1702 Ozaltin Oznur oznurozaltin@hacettepe.edu.tr Yeniay Ozgur grid.14442.37 0000 0001 2342 7339 Department of Statistics, Institute of Science, Hacettepe University, Beytepe, Ankara, 06800 Turkey 15 12 2022 120 3 12 2022 © The Author(s), under exclusive licence to Springer-Verlag GmbH Germany, part of Springer Nature 2022, Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. Nowadays, the number of sudden deaths due to heart disease is increasing with the coronavirus pandemic. Therefore, automatic classification of electrocardiogram (ECG) signals is crucial for diagnosis and treatment. Thanks to deep learning algorithms, classification can be performed without manual feature extraction. In this study, we propose a novel convolutional neural networks (CNN) architecture to detect ECG types. In addition, the proposed CNN can automatically extract features from images. Here, we classify a real ECG dataset using our proposed CNN which includes 34 layers. While this dataset is one-dimensional signals, these are transformed into images (scalograms) using continuous wavelet transform (CWT). In addition, the proposed CNN is compared to known architectures: AlexNet and SqueezeNet for classifying ECG images, and we find it more effective than others. This study, which not only performed CWT but also implemented short-time Fourier transform, examines the success in recognizing ECG types for the proposed CNN. Besides, different split methods: training and testing, and cross-validation are applied in this study. Eventually, CWT and cross-validation are the best pre-processing and split methods for the proposed CNN, respectively. Although the results are quite good, we benefit from support vector machines (SVM) to obtain the best algorithm and for detecting ECG types. Essentially, the main aim of the study increases classification results. In this way, the proposed CNN is utilized as deep feature extractor and combined with SVM. As a conclusion of this study, we achieve the highest accuracy of 99.21% from the proposed CNN–SVM when using CWT. Therefore, we can express that this framework can be used as an aid to clinicians for ECG-type identification. Keywords Convolutional neural networks (CNN) Continuous wavelet transform (CWT) Feature extraction Scalogram Support vector machine (SVM) ==== Body pmcIntroduction The qualitative processing and classification of biomedical signals is very important for diagnosis and therapy. Many methods are used to process biomedical signals. Some important methods are discrete Fourier transform (DFT), short-time Fourier transform (STFT), continuous wavelet transform (CWT), and discrete wavelet transform. The Fourier transformation provides a very good frequency range for stationary signals (Haberl et al. 1989). However, the time domain is almost non-existent. This can lead to serious problems, especially if time-dependent characteristics are to be inferred. However, when signals are transformed with the wavelet transform, both frequency and time domains are distinguishable (Li et al. 1995). In other words, wavelet transform (WT) is a transformation technique that splits signals into different frequency components and processes each component with the time domain of the respective scale. In this study, we focus on electrocardiogram (ECG) signals. The signals resulting from the electrical activity of the heart, the main vital organ in the human body, are called an electrocardiogram (ECG). Sudden deaths from heart disease with coronavirus (COVID-19) are currently on the rise (https://www.chss.org.uk/media-release/new-nhs-figures-show-dangerous-domino-effect-of-pandemic-on-progress-made-with-strokes-and-heart-disease/). For this reason, the processing and analysis of the signals received by the heart are very important for rapid diagnosis and treatment. In conventional methods, a suitable sampling method is used in the pre-processing phase of ECG signals and the signals are cleaned of noise. Then, the manual feature extraction phase begins, where it is very important to seek expert opinions. This phase is very critical as incorrect feature extraction can lead to misclassification of signals and serious errors in diagnosis and treatment. After all these phases are completed, classification is done using traditional classification algorithms. However, the studies show that the situation for deep learning algorithms has changed in recent years (Ozaltin et al. 2022; Özaltın and Yeniay 2021; Koc et al. 2022). Thanks to deep learning algorithms, successful classifications can be made automatically. In this way, the state of health of patients can be monitored with smartphones, watches, etc., even without an expert opinion. The aim of the study was to recognize type of ECG efficiently via deep learning algorithm. Firstly, we collect the dataset from PhysioNet databases (Physionet 2020). The dataset consists of three different types: arrhythmia (ARR), congestive heart failure (CHF), and normal sinus rhythm (NSR). In this study, a novel convolutional neural networks (CNN) architecture, which is one of the deep learning algorithms, is proposed for automatic ECG signal classification. This newly proposed 34-layer CNN architecture is designed for two-dimensional images. In fact, the newly proposed CNN is considered not only ECG classification, but also other biomedical signals, images, etc. classification. In this context, the ECG signals are naturally transformed from one-dimensional signals into images by using a continuous wavelet transform (CWT) in the pre-processing phase. This wavelet transform has three different mother wavelet functions: Amor, Bump, and Morse, which are the most commonly used. The impact of these functions on classification performance is also examined. In this study, 360 Hz, 500 Hz and 1000 Hz sample lengths are examined whether the wave characteristics become more evident. Figure 1 shows the images (scalograms) obtained with different sampling lengths of ECG signals, 360 Hz, 500 Hz, and 1000 Hz, respectively. Therefore, a total of nine different datasets are obtained under these conditions. These datasets are classified separately with the same training options parameters using the proposed CNN, AlexNet, and SqueezeNet. After identifying the best wavelet function, sample length, and architecture, we additionally investigate another pre-processing method: STFT to measure ECG classification performance via different split methods: training and testing, and cross-validation. Finally, the proposed CNN is used as a deep feature extractor from images and merged with support vector machines (SVM) to get trusted results.Fig. 1 227 × 227 × 3 size of images with different sample lengths In this study, a hybrid algorithm is proposed to detect ECG types from acquired images based on a deep learning algorithm and a machine learning algorithm. The main contributions and novelties of this study are as follows:When using CWT, 500 Hz is observed as an efficient sample length while converting. Amor wavelet function has higher performance than others while applying CWT. A new CNN architecture called proposed CNN is presented and compared with AlexNet and SqueezeNet. Eventually, the proposed CNN has the highest performance. To measure the performance of the proposed CNN, STFT is also used as pre-processing method via different splitting methods: training and testing (80:20, 70:30), and k-fold cross-validation (5, 10). Finally, CWT is higher than it and cross-validation is the best splitting method. To improve classification performance, the proposed CNN is utilized as feature extractor and benefited from both fully connected layer and maximum pooling layer. Reduced features are classified using SVM. Consequently, the highest performance to recognize ECG types is acquired thanks to the proposed CNN–SVM hybrid algorithm. Related studies Nowadays, artificial intelligence is evolving day by day, and many studies are also being conducted to classify ECG signals and other biomedical signals using CNN architectures. Khorrami and Moavenian (2010) applied the CWT, discrete wavelet transform (DWT), and discrete cosine transform (DCT) to ECG signals. In addition, they compared SVM with multi-layer perceptron (MLP) algorithms in the classification phase. In particular, they found that combinations made with MLP (CWT-MLP, DWT-MLP, DCT-MLP) are superior to SVM. Al Rahhal et al. (2018) transformed signals from different datasets using CWT to identify arrhythmias in ECG signals. Also, they used the CNN algorithm and achieved an accuracy of 99% in the classification phase. Huang et al. (2019) converted ECG signals with STFT and obtained two-dimensional scalograms in their study. Moreover, they benefited from the CNN architecture for classifying these scalograms and achieved an accuracy of 99%. In addition, they also classified the one-dimensional ECG signals using CNN and found an accuracy of 90.93%. Krak et al. (2020) transformed ECG signals into the images using CWT and DWT in their study. Furthermore, they classified the images using the CNN architecture and obtained an accuracy of 96% in the classification phase. Baloglu et al. (2019) designed a 10-layer end-to-end CNN architecture for the classification of multiclass one-dimensional ECG data and achieved an accuracy of a 99.78%. Mahmud et al. (2020) created a CNN architecture for multiclass one-dimensional ECG data and obtained an accuracy rate of 99.28%. Salem et al. (2018) utilized DenseNet architecture to classify transformed two-dimensional ECG data and achieved an accuracy of 97.23%. Zhao et al. (2020) proposed a CNN containing 24 layers for classifying transformed ECG data and achieved an accuracy of 87.1%. Xu and Liu (2020) created a CNN architecture in order to analyze ECG data recorded from a Holter device and achieved an accuracy of 99.4%. Rajkumar et al. (2019) suggested a CNN architecture for one-dimensional ECG data by using exponential linear unit (ELU) activation layers and achieved an accuracy of 93.6%. Hua et al. (2020) developed a CNN architecture for one-dimensional ECG signals and achieved an accuracy of 97.45%. Kiranyaz et al. (2015) proposed a CNN architecture for patient-specific real-time one-dimensional ECG classification and achieved an accuracy of 96.4%. Chen et al. (2020) suggested CNN + long short-term memory (LSTM) which can classify six kinds of ECG fragments. They have classified two ECG databases: MIT-BIH arrhythmia database and MIT-BIH arrhythmia database + Challenge2017, and achieved an accuracy of 99.32% and 97.15%, respectively, using CNN + LSTM. Sandeep et al. (2019) utilized the CNN architecture to classify ECG data and also achieved an accuracy of 90.63%. Furthermore, machine learning algorithms such as support vectors machine (SVM), K-nearest neighbors (KNN), decision tree (DT), extreme learning machine (ELM), ensemble learning, and multi-layer perceptron (MLP) to classify ECG signals by many other researchers (Alickovic and Subasi 2015; Qaisar and Subasi 2020; Tuncer et al. 2022; Ceylan and Özbay 2007; Pławiak and Acharya 2020). Additionally, Table 1 shows recent studies on ECG signals classification.Table 1 Recent Studies on ECG signals classification References Dataset Classes Methods Metrics Xing et al. (2022) MIT-BIH 5 classes SNN (Spiking Neural Network) Accuracy 98.26% Arrhythmia Dataset Sensitivity 94.75% F1-score 89.09% Chen et al. (2022) KMUH Dataset 9 classes DNN (Deep Neural Network) Accuracy 96.02% (Kaohsiung Medical University Hospital) Micro-F1 82.71% Pałczyński et al. (2022) PTB (Physikalisch-Technische Bundesanstalt) Dataset 2 classes CNN (1D) Accuracy 90.4% Sensitivity 89.6% F1-score 89.9% 5 classes FSL (Few-Shot Learning + SVM Accuracy 79.0% Sensitivity 70.6% F1-score 70.6% 20 classes CNN (1D) Accuracy 67.1% Sensitivity 32.4% F1-score 32.6% Cheng et al. (2022) MIT-BIH 5 classes U-net (1D) Accuracy 95.5% Arrhythmia Dataset Sensitivity 95.55% Specificity 97.64% Jiao and Wu (2022) MIT-BIH 5 classes Capsule network Accuracy 99.3% Arrhythmia dataset (LSTM + CNN) Sensitivity 83.1% F1-score 86.2% Sepahvand and Mohammadi (2022) Chapman ECG dataset 7 classes CNN Accuracy 98.15% (Chapman University and Shaoxing People’s Hospital) Sensitivity 97.11% F1-score 97.55% Eltrass et al. (2022) MIT-BIH 3 classes CNN (AlexNet) Accuracy 98.74% Arrhythmia dataset Sensitivity 98.17% Specificity 99.00% MIT-BIH normal sinus rhythm dataset BIDMC dataset Kumar et al. (2022) Collected ECG signals based on IoT (Internet of things) 2 classes Coy-Grey Wolf optimization + CNN Accuracy 95.05% Sensitivity 94.639% Specificity 94.639% Meng et al. (2022) MIT-BIH 3 classes Transformer with LightConv attention + CNN embedding Accuracy 99.32% Arrhythmia dataset The rest of the study is organized as follows: In Section 2, we present the materials and methods. Then, we explain the dataset, experimental setup, performance metrics, and experimental results in Section 3. Next, we discuss the results in Section 4. Finally, we conclude the study and state the future works. Materials and methods In this section, we first present pre-processing methods. Next, we introduce CNN, the proposed CNN, and pre-trained architectures: AlexNet (Krizhevsky et al. 2012) and SqueezeNet (Iandola et al. 2016). In the last, we present SVM and the proposed CNN–SVM architecture for classification of ECG dataset. Figure 2 shows the framework of this study.Fig. 2 Flowchart of this study Pre-processing methods In this study, we propose a novel CNN it needs images; therefore, we transform one-dimensional signals into two-dimensional image datasets via continuous wavelet transform (CWT) and short-time Fourier transform (STFT). Max–min normalization In this study, firstly, we normalize raw one-dimensional ECG signals using the minimum–maximum normalization method given formula in Eq. (1) as follows:1 X=signal-min(signal)max(signal)-min(signal) where X denotes the normalized ECG signal. Besides, min(.) is a minimum function, and max(.) is a maximum function. Continuous wavelet transform Continuous wavelet transform (CWT) is a transformation method. CWT allows simple analysis of its frequency components and can transform a one-dimensional signal into a two-dimensional scalogram by providing a mapping of the signal also on the time axis. The mathematical formulation of the CWT and WT family is offered in Eq. (2) and Eq. (3), respectively,2 CWTa,b=f,ψa,b∗=∫-∞+∞ftψa,b∗(t)dt 3 ψa,bt=1aψt-ba where f(t) is a continuous signal function received in this study as an ECG signal function, ψa,b(t) is the mother wavelet function, a indicates a scale parameter, b indicates the shift parameter or translation, and the symbol of * indicates the complex conjugate function (Lee and Choi 2019). Besides,f,ψa,b is expressed as a function of the inner products of Eq. (2). It CWTa,b is regulated,4 CWTa,b=1a∫-∞+∞f(t)ψt-badt will be in the form like in Eq. (4). The signal function f(t) can be converted from the inverse of CWTa,b, as follows:5 ft=1C∫-∞+∞∫-∞+∞CWTa,bψa,bta3/2dadb where C indicates the normalization constant depending on the choice of the mother wavelet function in Eq. (5) (Lee and Choi 2019). Some mother wavelet functions as follows:6 ψMorlt=e2πite-t22σ2=cos2πt+isin2πte-t22σ2 7 ψMexht=1-t2σ2e-t22σ2 8 ψBumpab=e1-11-ab-μ/σ2χμ-σ,μ+σ will be in the form in Eqs. (6–8). Here, ψMorlt, Morlet, ψMexht, Mexican hat, and ψBumpab, Bump, show the mother wavelet function (Lee and Choi 2019). Short-time Fourier transform (STFT) Short-time Fourier transform (STFT) is also a transformation method. The STFT is obtained from the discrete Fourier transform (DFT), to discover the sudden frequency and the sudden amplitude of localized waves with time-varying typical (Huang et al. 2019; Haykin and Veen 1999). The STFT uses a window function to extract time-domain information (Toma and Choi 2022). The window function possesses a certain interval, and the value of this window function outward of the interval is zero (Toma and Choi 2022). To calculate the frequency domain information, the window function shifts over all non-stationary signals and each time it is multiplied with the signal (Haykin and Veen 1999; Toma and Choi 2022). Further, the time–frequency spectrogram can be computed in a discretized non-stationary digital signal as given in Eq. (9) (Toma and Choi 2022),9 STFTxn=Xm,ω=∑n=-∞∞xnwn-me-jωn where xn symbolizes signals and wn is the window function. In this study, we utilize the Kaiser function with a window size of 500 Hz. Thus, we convert ECG signals into ECG spectrums images with dimensions of 227 × 227 × 3. Convolutional neural network (CNN) Convolutional neural network (CNN) emerges as a specialized deep learning approach for analyzing two-dimensional data. Not only it is preferred algorithm in the analysis of multidimensional data but also one-dimensional data. Other classifications and clustering algorithms are difficult to apply to real-time data due to their computational complexity (Narin 2020). For this reason, deep learning technology that can overcome this complexity evolves day by day. Moreover, CNN can perform feature extraction and classification automatically using raw data, so deep learning algorithms are very popular in the field of artificial intelligence. Further, it is found to give very good results of classification studies involving both big data and small data by researchers. Thanks to the CNN algorithm, ECG signals can be analyzed and observed on smartphones, watches, Holter monitoring devices, etc. (Huang et al. 2019). The CNN processes an image in different layers and separates all its features. The most commonly used layers are:Convolution layer, Nonlinear layer, Pooling layer, Flattening layer, Fully connected layer expressed as (Baloglu et al. 2019; Lee and Choi 2019; Acharya et al. 2017). Convolutional Layer: The convolution process is the layer where the features of the image are determined. To determine more than one feature, the number of convolutional layers increases in the same proportion. This layer is the main building block of CNN. Nonlinear Layer: This layer is also known as the activation layer. It is used to realize the activation of the system with nonlinear functions. Rectified linear unit function (ReLU), which is widely used because it is faster than others, is preferred in recent years. Pooling Layer: Smaller matrices are obtained while preserving the properties of the existing input. In this way, the computational complexity is reduced. Flattening Layer: The matrix format data obtained from the previous step is prepared following the fully connected layer. Fully Connected Layer: It is the most important layer of convolutional neural network layers. The data are taken from the flattening layer and trained by the neural network and the learning process is performed. Pre-trained architectures: AlexNet and SqueezeNet AlexNet (Krizhevsky et al. 2012) has five convolution layers combined with max-pooling layers and three fully connected layers. It also includes a dropout layer and a softmax. Moreover, each layer is activated with the ReLU activation function. In 2012, it was used the ReLU activation function in place of the tanh function (Abdelmalek et al. 2019). Thus, it was seen that the architecture was accelerated. The total number of parameters is 62.3 million, and the input image size is 227 × 227. SqueezeNet (Iandola et al. 2016) starts with an independent convolutional layer (conv1), follows by eight firing modules, and ends with the last convolutional layer (conv10). In total, it consists of ten convolutional layers, some max-pooling layers, and a SoftMax layer, in the recently presented version. In this study, a novel CNN architecture is presented in the next section and it is compared with AlexNet and SqueezeNet on created different datasets. Novel proposed CNN architecture A CNN architecture usually consists of an input layer, some convolutional layers, some pooling layers, and a fully connected layer (Krak et al. 2020). In this study, we introduce a novel CNN architecture. It has seven convolutional layers, seven batch normalization layers, seven activation layers (ReLU), seven maximum pooling layers, and two fully connected layers with one dropout layer. Additionally, a SoftMax layer and a classification layer with an entropy approach are used as well. The convolution layers are effectively utilized for feature extraction from ECG image datasets. This is important since well feature extraction is also meaning very sensitive classification. Essentially, these layers are filtered to enhance the features of the primary signal while reducing the noise (Hua, et al. 2020; Li et al. 2018). The pooling layers reduce the dimension of the input images, and these are prepared for the next layer. Finally, extensive features in the fully connected layers are reduced with 0.5 probability by using the dropout layer and transferred to the SoftMax layer for the classification. Details of the parameters of the proposed CNN are given in Table 2.Table 2 Proposed CNN architecture details Layer name Type Layer parameters Output shape Input Image Input 227 × 227 × 3 images with “zerocenter” normalization 227 × 227 × 3 Conv-1 Convolution 2D Filter size = 64, number of filters = [5 5], stride = [1 1], padding = [1 1 1 1], BatchNormalization, ReLU 225 × 225 × 64 MaxPool-1 Max Pooling Pool size = [3 3], stride = [2 2], padding = [0 0 0 0] 112 × 112 × 64 Conv-2 Convolution 2D Filter size = 128, number of filters = [3 3], stride = [1 1], padding = [1 1 1 1], BatchNormalization, ReLU 112 × 112 × 128 MaxPool-2 Max Pooling Pool size = [3 3], stride = [2 2], padding = [0 0 0 0] 55 × 55 × 128 Conv-3 Convolution 2D Filter size = 128, number of filters = [13 13], stride = [1 1], padding = [0 0 0 0], BatchNormalization, ReLU 55 × 55 × 128 MaxPool-3 Max Pooling Pool size = [3 3], stride = [2 2], padding = [0 0 0 0] 27 × 27 × 128 Conv-4 Convolution 2D Filter size = 256, number of filters = [7 7], stride = [1 1], padding = [1 1 1 1], BatchNormalization, ReLU 27 × 27 × 256 MaxPool-4 Max Pooling Pool size = [2 2], stride = [2 2], padding = [0 0 0 0] 13 × 13 × 256 Conv-5 Convolution 2D Filter size = 128, number of filters = [3 3], stride = [1 1], padding = [1 1 1 1], BatchNormalization, ReLU 13 × 13 × 128 MaxPool-5 Max Pooling Pool size = [3 3], stride = [2 2], padding = [0 0 0 0] 6 × 6 × 128 Conv-6 Convolution 2D Filter size = 128, number of filters = [3 3], stride = [1 1], padding = [1 1 1 1], BatchNormalization, ReLU 6 × 6 × 128 MaxPool-6 Max Pooling Pool size = [3 3], stride = [2 2], padding = [0 0 0 0] 3 × 3 × 128 Conv-7 Convolution 2D Filter size = 128, number of filters = [3 3], stride = [1 1], padding = [1 1 1 1], BatchNormalization, ReLU 3 × 3 × 128 MaxPool-7 Max Pooling Pool size = [2 2], stride = [2 2], padding = [0 0 0 0] 1 × 1 × 128 FC-8 Fully Connected 4096 1 × 1 × 4096 Drop-8 Dropout 50% FC-9 Fully connected 3 (number of classes) 1 × 1 × 3 Softmax SoftMax 1 × 1 × 3 Output Classification Cross-entropy The proposed CNN is a novel architecture that has different filter sizes, number of filters, strides, and padding. Fundamentally, we develop the architecture for biomedical image classification. However, it is tested on known classical datasets such as CIFAR-10, like other CNN architectures. Additionally, it is utilized on Physikalisch-Technische Bundesanstalt (PTB) Diagnostic ECG Database (Özaltın and Yeniay 2021; Goldberger, et al. 2000). This proposed CNN is performed for not only signals but also brain computed tomography, detailed in Ozaltin et al. (2022). Moreover, this proposed CNN is named as OzNet in studies of Ozaltin et al. (2022). And, this architecture obtains successful performances in these datasets (Fig. 3). Fig. 3 Transformed images using STFT In this study, the proposed CNN is compared with AlexNet and SqueezeNet using same fine-tuning parameters. Stochastic gradient descent method (sgdm) is performed as the optimization algorithm, and the momentum parameter is determined as 0.95, and the learning rate is also started with 0.0001 as constant. Figure 4 shows the proposed CNN scheme.Fig. 4 Proposed CNN architecture Deep feature extraction In this study, the proposed CNN can extract features from images effectively. Therefore, we use it both classifier and deep feature extractor. Although, when it is used for classification algorithm, the results are quite well, we decide to more improving results for obtaining the best one. Therefore, we designed a hybrid algorithm which is included the proposed CNN and SVM. In this section of study, the proposed CNN is assigned as automatic feature extractor from ECG images and SVM is employed for classifier. In brief, we can explain the steps of how to work it as follows: (i) the proposed CNN is trained on ECG images, firstly. (ii) Reduced features are obtained from the proposed CNN of fully connected layer and 4096 features are collected for each image. (iii) To classify with these features, the dataset is split into 30% training set and 70% testing set. This is because we want to obtain trustworthy classification results owing to dropout layer would not have much influence (Elleuch et al. 2016; Srivastava et al. 2014). Then, the trained net is activated. (iv) SVM classifier is employed to detect type of ECG, effectively. The same stages are happened when reduced features are achieved from maximum pooling (Max-Pooling 7) layer. Figure 5 demonstrates the scheme of the proposed CNN–SVM.Fig. 5 Proposed CNN–SVM algorithm Support vector machine (SVM) Support vector machine (SVM) is a machine learning algorithm that an effective separation with a kernel-based method to the datasets for classification or regression (Koklu and Ozkan 2020). It is improved by Cortes and Vapnik (1995) for two classes. Then, the algorithm is advanced and generalized for multiclass and nonlinear datasets. In general, the dataset can be separated in high-dimensional feature space with a kernel function. Also, SVM can be overcome confused datasets and overfitting. The most common representation of the SVM function is f(x)=wTϕx+b where w∈Rn b∈R and ϕx is a feature map. Results ECG dataset In this study, we benefit from three different ECG datasets from PhysioNet databases (Physionet 2020). Each raw ECG dataset is taken with a signal length of 1 h and sampled at 128 Hz. The first ECG dataset consists of the ECG recordings from 48 patients, which contain two leads. It is received from the MIT-BIH Arrhythmia Database and referred to as ARR (Goldberger, et al. 2000; Moody and Mark 2001). The next ECG dataset consists of the ECG recordings from 15 patients, which contain two leads. It comes from the BIDMC Congestive Heart Failure Database and is named CHF (Goldberger, et al. 2000; Baim et al. 1986). The final ECG dataset consists of the ECG recordings from 18 patients, containing two leads. It is obtained from MIT-BIH Normal Sinus Rhythm and referred to as NSR (Goldberger, et al. 2000). There are a total of 96 ARR, 30 CHF and 36 NSR in the ECG dataset. In fact, this dataset is not suitable for convolutional neural networks because of demand pattern. That is why we convert the signals into the images. First, we normalize the dataset using the max–min normalization method. Next, one-dimensional ECG signals are transformed into images utilizing CWT with different sampling lengths of signals, 360 Hz, 500 Hz, and 1000 Hz. This is because we want to compare which sample length is better to see differences. Besides, three different mother wavelet functions: Amor, Bump, and Morse, are applied to each sample length to compare which mother wavelet function is better to detect differences. It also sizes each image to 227 × 227 × 3 and.jpg format. Therefore, we create nine different balanced datasets with identifying mother wavelet functions and signal lengths. Each dataset contains 900 images, and each class (ARR, CHF, and NSR) includes 300 images. After that, to compare the results, we also benefit from the STFT transform method to turn signals into images. Also, created this dataset consists of 900 images, and each class contains 300 images. Experimental setup In this study, we run AlexNet, SqueezeNet, and the proposed CNN to classify ECG datasets. In this study, we use splitting methods: training and testing sets, and cross-validation to compare affective classification performance. Primarily, the dataset is split conventionally as a training and testing set with 80:20 and 70:30 percentages. Next, k-fold cross-validation is performed, where k values are determined as 5 and 10. Further, we use the proposed CNN to automatically extract deep features. They are reached from the fully connected layer (FC-8) and maximum pooling layer (Max-Pooling 7), respectively. To classify these reduced features, we perform an SVM using Gaussian kernel function to detect ECG type from images. Therefore, we present a comprehensive study that effectively determines the ECG type. Performance metrics In this study, we review performance metrics of CNN architectures that are accuracy, sensitivity, specificity, precision, and F1-score in Eq. (8–12), as follows (Xu and Liu 2020; Abdelmalek et al. 2019):10 Accuracy=TP+TNTP+TN+FP+FN×100% 11 Sensitivity=TPTP+FN×100% 12 Specificity=TNTN+FP×100% 13 Precision=TPTP+FP×100% 14 F1 - Score=2×Precision×SensitivityPrecision+Sensitivity×100% where TP: true positive, FP: false positive, TN: true negative, and FN: false negative are expressed. Experimental results This study is conducted in a MATLAB 2021b environment with Intel Core i7-7500U CPU, NVIDIA GeForce GTX 950 M, 16 GB RAM and 64-bit Operating System. The aim of this study was to identify ECG types via CNN architectures and a designed hybrid algorithm. First, nine different ECG image datasets are created using CWT, and each is classified using AlexNet, SqueezeNet, and the proposed CNN with the same option parameters with 80:20 training and testing split percentages. In addition, the obtained results are tested with the Wilcoxon signed rank test. Tables 3 and 4 show both performance results and paired comparisons for statistical significance. Besides, all comparisons are demonstrated in Fig. 5.Table 3 Wilcoxon signed rank test for proposed CNN and AlexNet Datasets AlexNet (accuracy%) Proposed CNN (accuracy%) Di rDi ECG signal length Wavelet function 360 Hz Amor 89.33 92.67 −3.34 7 Bump 91.33 92 −0.67 1 Morse 94 94 0 – 500 Hz Amor 94.67 98 −3.33 6 Bump 94 94 0 – Morse 93.33 94.67 −1.34 2.5 1000 Hz Amor 93.33 95.33 −2 4.5 Bump 92 94 −2 4.5 Morse 93.33 94.67 −1.34 2.5 Wilcoxon signed ranks test Z=-2.375,p-value=0.018 ∑i=1n=7rDi=28 Table 4 Wilcoxon signed rank test for proposed CNN and SqueezeNet Datasets SqueezeNet (accuracy%) Proposed CNN (accuracy%) Di rDi ECG signal length Wavelet function 360 Hz Amor 90 92.67 −2.67 4 Bump 94 92  + 2 2.5 Morse 89.33 94 −4.67 6.5 500 Hz Amor 87.33 98 −10.67 9 Bump 94.67 94  + 0.67 1 Morse 90.67 94.67 −4 5 1000 Hz Amor 88.67 95.33 −6.66 8 Bump 92 94 −2 2.5 Morse 90 94.67 −4.67 6.5 Wilcoxon signed ranks test Z=-2.255,p-value=0.024 ∑i=1n=7r-Di=41.5∑i=1n=2r+Di=3.5 When Table 3 is examined in relation to the sample length of the ECG between AlexNet and the proposed CNN, AlexNet gets a maximum accuracy of 94.67% at a sample length of 500 Hz. Also, the proposed CNN achieves the maximum accuracy of 98.00% at a sample length of 500 Hz. Finally, SqueezeNet achieves a maximum accuracy of 94.67% with a sample length of 500 Hz, as given in Table 4. Therefore, we can indicate that 500 Hz is the best one for the sample length of ECG. When Tables 3 and 4 are also examined in terms of the mother wavelet function, Amor and Morse provide almost similar results to classify images for AlexNet and our proposed CNN. However, these results do not apply to SqueezeNet. When SqueezeNet is examined for the mother wavelet function, Bump is found to be the best. So, if researchers want to use SqueezeNet, they can choose to use the bump wavelet function while performing CWT. When Tables 3 or 4 is investigated for the proposed CNN in terms of the mother wavelet function, Amor’s choice for classifying the images is the best. Although the results are quite good, we want to test these results for the reliability of this study using the nonparametric method, the Wilcoxon signed rank test. First, we make one hypothesis, which is a null hypothesis: there is no difference between AlexNet and the proposed CNN, and an alternative hypothesis: there is a difference between AlexNet and the proposed CNN. As a result, p value is obtained 0.018 < 0.05, and hence, null hypothesis is rejected. In this study, a significant level is determined as 0.05. Therefore, we can statistically say that there is a difference between AlexNet and the proposed CNN. Though the results are rather good, as given in Table 4, we want to test these results for the trustfully of this study using the Wilcoxon signed rank test. First, we make one hypothesis, which is a null hypothesis: there is no difference between SqueezeNet and the proposed CNN, and an alternative hypothesis: there is a difference between SqueezeNet and the proposed CNN. As a result, p value is obtained 0.024 < 0.05, and hence, null hypothesis is rejected. Thus, we can statistically express that there is a difference between SqueezeNet and the proposed CNN. As a result, the proposed CNN is the best choice to classify ECG datasets while using CWT and 80:20 training and testing percentages. Figures 6 and 7 display performance graphs for classification. In addition, Table 5 details the results with other performance metrics for each class.Fig. 6 Performance comparison of different sample lengths and mother wavelet function using CNN architectures Fig. 7 Performance comparison of different sampling lengths and mother wavelet function using the proposed CNN Table 5 Performance metrics of proposed CNN, AlexNet, and SqueezeNet architectures CNN architecture Class name Sensitivity (%) Specificity (%) Precision (%) F1-score (%) Test Accuracy Rate (%) Proposed-CNNa ARR 97.96 98.02 96 96.97 98 CHF 98 99.02 98 98 98 NSR 98.04 100 100 99.01 98 Proposed-CNNb ARR 93.88 96.04 92 92.93 95.33 CHF 96.15 100 100 98.04 95.33 NSR 95.92 98.01 94 94.95 95.33 AlexNeta ARR 92.31 97.96 96 94.12 94.67 CHF 97.87 96.12 92 94.85 94.67 NSR 94.12 98.6 96 95.05 94.67 SqueezeNeta ARR 94.34 100 100 97.09 94.67 CHF 90.57 97.94 96 93.20 94.67 NSR 100 96.05 88 93.62 94.67 aECG sample length 500 Hz, bECG sample length 1000 Hz *Bold values indicate the maximum metrics When all of the performance metrics in Table 5 are examined, these proposed CNNs metrics are met at over 96%. Specifically, the NSR performances are considered to be %100 in terms of specificity and precision score. In addition, its performances on other metrics are also over 98%. When the metrics are examined, which the classifiers did well, it is noticeable that the proposed CNN’s F1-score is superior to the others in Table 5. Therefore, the proposed CNN is determined to be the best classifier in terms of performance metrics. As a result of this part, the best signal length, mother wavelet function, and architecture are determined to be 500 Hz, Amor, and the proposed CNN, respectively. Thus, these foundations have shown that only one ECG dataset is classified. In addition, Fig. 8 shows the accuracy rate graph and the loss graph for the proposed CNN, while the signal length is 500 Hz and the wavelet function is Amor.Fig. 8 Accuracy rate and loss graph of training progress using the proposed CNN Having determined the proposed CNN as the best architecture for classifying ECG images, we examine the impact of other split methods on performance. First, the ECG image dataset created with 500 Hz sample length and Amor wavelet function using CWT is divided into 80:20 and 70:30 training and test sets, respectively, and then, we use a fivefold and tenfold cross-validation. The results are shown in Tables 6, 7, and 8.Table 6 Proposed CNN performance metrics over five training sessions with an 80:20 training and testing split using CWT Training number Sensitivity (%) Specificity (%) Precision (%) F1-score (%) Test accuracy rate (%) 1 94.86 97.60 94.67 94.66 94.67 2 95.53 97.72 95.33 95.33 95.33 3 96.05 98.12 96.00 96.00 96 4 98.06 99.03 98.01 97.99 98 5 98.68 99.45 98.67 98.67 98.67 Mean + Std 96.64 ± 1.65 98.38 ± 0.82 96.54 ± 1.72 96.53 ± 1.73 96.53 ± 1.73 Table 7 Proposed CNN performance metrics over five training sessions with a 70:30 training and testing split using CWT Training number Sensitivity (%) Specificity (%) Precision (%) F1-score (%) Test accuracy rate (%) 1 92.96 96.48 93.49 92.93 92.96 2 94.44 97.22 94.57 94.43 94.44 3 96.67 98.33 96.70 96.66 96.67 4 95.92 97.96 96.24 95.96 95.93 5 97.04 98.51 97.06 97.04 97.04 Mean + Std 95.41 ± 1.69 97.70 ± 0.84 95.61 ± 1.52 95.4 ± 1.705 95.41 ± 1.69 Table 8 Proposed CNN performance metrics with fivefold and tenfold cross-validation using CWT k Sensitivity (%) Specificity (%) Precision (%) F1-score (%) Accuracy rate (%) 5 96.44 98.22 96.44 96.44 96.44 10 97.22 98.61 97.23 97.22 97.22 *Bold values indicate the maximum average metrics When Table 6 is viewed, all mean performance metrics are observed above 96.52% and also the maximum standard deviation (Std) was 0.0173. Therefore, the proposed architecture is traditionally trained and tested to classify images. According to Table 7, all mean performance metrics are above 95.3% and also the maximum standard deviation (Std) was 0.01705. Thus, it can be said that a training and testing split of 80:20 has the best performance for classifying ECG images while performing CWT. According to Table 8, all average performance metrics are seen, and the maximum average accuracy of 97.22% is obtained through tenfold cross-validation. Concluding on the use of CWT, the performances expressed that the cross-validation is better than the split method for training and testing. Perfect performances for classifying ECG images are achieved using CWT and the proposed CNN. However, we would like to see how other pre-processing methods affect the performance of the proposed CNN using the same splitting methods. Therefore, we prefer to use STFT method which is performed widely. Its performances are shown in Tables 9, 10, and 11.Table 9 Proposed CNN performance metrics over five training sessions with an 80:20 training and testing split using STFT Training number Sensitivity (%) Specificity (%) Precision (%) F1-score (%) Test accuracy rate (%) 1 87.78 93.89 87.74 87.73 87.78 2 90.55 95.27 90.69 90.57 90.56 3 88.89 94.44 89.15 88.89 88.89 4 89.44 94.72 90.28 89.27 89.44 5 90.56 95.28 91.04 90.29 90.56 Mean + Std 89.44 ± 1.177 94.72 ± 0.588 89.78 ± 1.3438 89.35 ± 1.142 89.45 ± 1.18 Table 10 Proposed CNN performance metrics over five training sessions with a 70:30 training and testing split using STFT Training number Sensitivity (%) Specificity (%) Precision (%) F1-score (%) Test accuracy rate (%) 1 87.04 93.52 88.02 86.72 87.04 2 91.48 95.74 91.77 91.43 91.48 3 89.25 94.62 90.05 89.14 89.26 4 89.25 94.63 89.34 89.23 89.26 5 89.63 94.81 89.95 89.71 89.63 Mean + Std 89.33 ± 1.58 94.66 ± 0.789 89.83 ± 1.35 89.25 ± 1.69 89.33 ± 1.578 Table 11 Proposed CNN performance metrics with fivefold and tenfold cross-validation using STFT k Sensitivity (%) Specificity (%) Precision (%) F1-score (%) Accuracy rate (%) 5 91.11 95.56 91.16 91.06 91.11 10 87.66 93.83 87.87 87.57 87.66 According to Table 9, all the average performance metrics are shown above 89.3% and also, the maximum standard deviation (Std) was 0.013438. Therefore, the proposed architecture is traditionally trained and tested at 80:20 to classify images using STFT. According to Table 10, all the average performance metrics are shown above 89.2% and also, the maximum standard deviation (Std) was 0.0169. Therefore, when the proposed architecture is trained and tested at 70:30 to classify images using STFT, performance results are similar to 80:20 training and testing split. According to Table 11, all average performance metrics are observed, and the maximum average accuracy of 91.11% is achieved through fivefold cross-validation. Final on the use of STFT, the performances indicated that the cross-validation is better than the split method for training and testing. Compared with CWT, STFT is not preferred to create ECG images as its performances are lower than CWT using the proposed CNN. In general, however, the proposed CNN in this study achieves quite good classification performance for recognizing ECG types. Indeed, in this study, our main contributor wants to find the best algorithm to detect ECG types. Thus, the proposed CNN is used as a deep feature extractor from images. Having trained proposed CNN for the ECG images using CWT through 80:20 splitting method because of the highest accuracy rate, reduced features are obtained from the fully connected (FC-8) layer and maximum pooling layer (Max-Pooling 7), respectively. These features are classified using SVM classifier. Therefore, we designed novel hybrid algorithm thanks to the proposed CNN and SVM. Table 12 exhibits performance results.Table 12 Performance metrics of proposed CNN–SVM algorithm Layer name Sensitivity (%) Specificity (%) Precision (%) F1-score (%) Test accuracy rate (%) Max Pooling-7 99.206 99.66 99.213 99.206 99.21 FC-8 98.72 99.5 98.75 98.732 98.73 *Bold values indicate the maximum metrics of this study According to Table 12, all performance metrics are increased for two different processes. However, the highest accuracy of 99.21% is achieved when retrieving features from Max-Pooling 7 layer. In this study, while using CWT, the proposed CNN–SVM is seen as the best algorithm for recognizing ECG types. Additionally, Fig. 9 displays a confusion matrix of the proposed CNN–SVM with the highest.Fig. 9 Confusion matrix of the proposed CNN–SVM This study is conducted not only with CNN, but also with an SVM classifier, which is very successful in image classification. The combination of these two methods, which are very successful individually, has proven itself very well. Table 13 shows a comparison of all methods in terms of performance metrics while using CWT.Table 13 Comparison of all methods in terms of performance metrics when using CWT Classification algorithm Sensitivity (%) Specificity (%) Precision (%) F1-score (%) Test accuracy rate (%) Proposed CNN 96.64 98.38 96.54 96.53 96.53 SVM 85.56 93.68 85.56 85.51 85.56 Proposed CNN–SVM 99.206 99.66 99.213 99.206 99.21 Discussion In this study, we aim to investigate whether ECG types are distinguishable from ECG-created images using deep learning structures and which type of ECG images (CWT or STFT) is efficient in recognizing ECG types using deep learning. Actually, our study possesses some advantages and disadvantages as follows: Advantages of this study are as follows: (i) Different sample lengths (360 Hz, 500 Hz, and 1000 Hz) are researched while using CWT, and 500 Hz is seen as an efficient sample length when one-dimensional signals are converted into images. (ii) Different mother wavelet functions (Amor, Morse, and Bump) are examined which one is more efficient on CNN architectures classification performance while performing CWT. (iii) This study presents a novel CNN architecture, called proposed CNN, and it is compared with AlexNet and SqueezeNet. (iv) Amor wavelet function is viewed successfully when using AlexNet and the proposed CNN, and the Bump wavelet function is high performance for SqueezeNet. (v) The proposed CNN has the highest performance in generating ECG datasets and is tested for significant differences via the Wilcoxon signed rank test. (vi) CWT is compared with the STFT method using the proposed CNN. (vii) Performances are measured on different splitting methods: training and testing (80:20, 70:30), and k-fold cross-validation (5, 10). (viii) The proposed CNN is performed as a deep feature extractor and provides from fully connected and maximum pooling layer. (ix) As a result, a new hybrid algorithm with the proposed CNN and SVM is designed. In this stage, SVM is used as a classifier to increase the performance of the distinguishability of ECG types. Disadvantages of this study are researched limited ECG types (ARR, CHF, and NSR) and the number of individuals. Many approaches are used for the classification of arrhythmia (ARR), congestive heart failure (CHF), and normal sinus rhythm (NSR) datasets. Basically, successful classification is very important for diagnosis and treatment. Therefore, in this study, we propose a novel 34-layer deep learning algorithm, called proposed CNN. Besides this ECG dataset, other datasets have also been classified using our proposed CNN, such as the PTB ECG dataset, CT images of brain hemorrhages, and the CIFAR-10 dataset. As is known, the pre-trained CNN architectures are tested on the traditional dataset. In addition, the proposed CNN architecture is also tested on the CIFAR-10 dataset in this study and examined whether it could make a successful classification. The CIFAR-10 dataset consists of 10 classes and 60,000 images. Similarly, this huge dataset is split 80% for training and 20% for testing, as shown in the study. In this way, 50,000 images are trained and 10,000 images are also tested. Also, the same option parameters are applied to both sets of data. Table 14 shows the proposed CNN success on different datasets. In addition, Fig. 10 displays the confusion matrix for the CIFAR-10 dataset.Table 14 The proposed CNN performance on different datasets Datasets Number of Class Sensitivity (%) Specificity (%) Precision (%) F1-score (%) Accuracy rate (%) PTB ECG dataset (Özaltın and Yeniay 2021) 2 96.42 94.96 95 95.56 95.6 CIFAR-10 10 83.95 98.22 84.10 83.87 84 ECG dataset in this Study 3 96.64 98.38 96.54 96.53 96.53 CT images of brain hemorrhage (Ozaltin et al. 2022) 4 91.95 94.93 93.17 92.53 92.85 Fig. 10 Confusion matrix of the proposed CNN for the CIFAR-10 dataset As can be seen, the performance of the proposed CNN is very good. However, as mentioned earlier, this CNN must be excellent for classifying biomedical signals or images. Therefore, the proposed CNN is merged with SVM for perfect classification. In general, if a CNN architecture has a fully connected layer, that layer is used for obtaining features and combined with SVM. Of course, this method offers good advantages because of the extracted features. However, the deep learning algorithm (also CNN) is a complex nonlinear model and is referred to as a black box (Guidotti et al. 2018). Accordingly, it has to be investigated which last layers have good properties within this probabilistic process. Among all these considerations, the characteristics in the Max-Pooling 7 (just before the FC-8 layer) are also examined in the present study. According to the knowledge gained in this study, it is necessary to examine the features in the last layers for a more sensitive analysis, which are listed in Table 12. Apart from this, when the literature is searched on the same property ECG dataset, the proposed CNN–SVM hits the top in terms of accuracy rate, detailed in Table 14. Conclusion Many of sudden deaths from heart disease continue to increase these days with the coronavirus (COVID-19). Based on this, the automatic classification of the signals received from the heart is of great importance for diagnosis and treatment. In this study, we classify ECG types using our proposed CNN, which has overcome overfitting with the dropout layer. This CNN is also performed on other datasets, shown in Table 14. In addition, the proposed CNN is compared to AlexNet and SqueezeNet on nine different ECG image datasets processed via CWT using three different wavelet functions and three different sample lengths. All results show that the best sample length is 500 Hz and the best mother wavelet function is “Amor.” Also, the comparison of classification success in terms of the overall accuracy rate of the proposed CNN, AlexNet, and SqueezeNet is 98%, 94.67%, and 94.67%, respectively. Therefore, the proposed CNN architecture performs the best classification on the ECG image dataset generated with the Amor wavelet function and the 500 Hz sample length by using CWT. However, we want to search how another pre-processing method affects classification success and so, we generate new ECG images using STFT with 500 Hz sample length. In this way, we use not only a splitting method as training and testing (80:20, 70:30), but also cross-validation implemented on two created datasets. According to the ECG image dataset generating via CWT, when the dataset split training and testing as 80:20, all mean performance metrics are over 96.5%, and also maximum standard deviation (Std) is 0.0173 on testing the ECG dataset. When the dataset split training and testing as 70:30, all average performance metrics are over 95.3%, and the highest Std is 0.01705. Further, as fivefold and tenfold cross-validation methods are implemented on the dataset, average accuracies are 96.44% and 97.22%, respectively. Also, the maximum average accuracy of 97.22% is obtained through tenfold cross-validation. Resulting of the use of CWT, the performances expressed that cross-validation is better than training and testing. According to the ECG image dataset creating via STFT, when the dataset split training and testing as 80:20, all average performance metrics are above 89.3% and also the maximum Std is 0.013438. While the dataset split training and testing as 70:30, all mean performance metrics are above 89.2% and also the maximum Std is 0.0169. Besides, when fivefold and tenfold cross-validation methods are applied on the dataset, average accuracies are 91.11% and 87.66%, respectively. All these results show that CWT is better than STFT to detect types of ECG. The main purpose of the study is to find an excellent classification algorithm for recognizing the ECG types. Therefore, the proposed CNN is merged with SVM. In this stage of the study, the proposed CNN is employed as a deep feature extractor from ECG images generated with CWT. In general, if any CNN architecture has a fully connected layer, it is used for obtaining features. It is highlighted that it can provide an advantage to examine features from the last layers of CNN, such as the max-pooling layer, in this study. To improve the proposed CNN performance, Max-Pooling 7 and FC-8 layers are used attaining reduced features, and the results are detailed in Table 12. As a result, the highest success with an accuracy of 99.21% is achieved by Max-Pooling 7 layer. When comparing to other studies on similar ECG datasets, the proposed CNN–SVM is considered the best performing for classification, detailed in Table 15.Table 15 The comparison of classification performances for different studies on ECG signals Study Pre-processing method Algorithm Accuracy (%) Çınar and Tuncer (2021) STFT CNN (AlexNet-SVM) 96.77 Eltras et al. (2021) CQ-NSGT* CNN (AlexNet) 98.82 Gaddam et al. (2021) CWT CNN (AlexNet) 95.67 Golgowski and Osowski (2020) CWT CNN 82.06 DWT Extra random forests 97.78 Krak et al. (2020) CWT CNN 96 Krishnakumar et al. (2021) CWT CNN (GoogleNet) 96.88 Kumari et al. (2020) DWT SVM 95.92 Nahak and Saha (2020) RR Wavelet with AR Fusion of features SVM SVM SVM 86.77 92.22 93.33 Olanrewaju et al. (2021) CWT CNN (AlexNet) 98.7 Rahuja and Valluru (2021) CWT CNN (AlexNet) 97.3 Proposed CNN CWT CNN 96.53 Proposed CNN–SVM CWT CNN–SVM 99.21 *Constant-Q non-stationary Gabor transform This study applies deep learning algorithms for ECG-type detection as an assisting decision support system. As such, clinicians will not spend much more time identifying ECG types, and the proposed pipeline will help physicians and professionals better identify ECG types in a hospital setting. In future work, we will continue to search for the detection of various diseases on signals or images by deep learning algorithms. Acknowledgements This study is based on Oznur Ozaltin’s Ph.D. thesis and supervised by Ozgur Yeniay. Author contributions OO idealized this study and analyzed the data. OY supervised the research and approved the final draft. Funding The authors took on no certain funding for this study. Data availability Data can be available from https://www.physionet.org/. Declarations Conflict of interest The authors announced that they had no conflicts of interest to report related to this study. Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References Abdelmalek B, Ahmed K, Amine TM (2019) Lightweight CNNs-Based Object Detection forEmbedded Systems implementation. In: 1st Conference on innovative trends in computer science (CITCS'2019) Acharya UR Fujita H Oh SL Hagiwara Y Tan JH Adam M Application of deep convolutional neural network for automated detection of myocardial infarction using ECG signals Inf Sci 2017 415 190 198 10.1016/j.ins.2017.06.027 Al Rahhal MM Bazi Y Al Zuair M Othman E BenJdira B Convolutional neural networks for electrocardiogram classification J Med Biol Eng 2018 38 6 1014 1025 10.1007/s40846-018-0389-7 Alickovic E Subasi A Effect of multiscale PCA de-noising in ECG beat classification for diagnosis of cardiovascular diseases Circuits Syst Signal Process 2015 34 2 513 533 10.1007/s00034-014-9864-8 Baim DS Survival of patients with severe congestive heart failure treated with oral milrinone J Am Coll Cardiol 1986 7 3 661 670 10.1016/S0735-1097(86)80478-8 3950244 Baloglu UB Talo M Yildirim O San Tan R Acharya UR Classification of myocardial infarction with multi-lead ECG signals and deep CNN Pattern Recognit Lett 2019 122 23 30 10.1016/j.patrec.2019.02.016 Ceylan R Özbay Y Comparison of FCM, PCA and WT techniques for classification ECG arrhythmias using artificial neural network Expert Syst Appl 2007 33 2 286 295 10.1016/j.eswa.2006.05.014 Chen C Hua Z Zhang R Liu G Wen W Automated arrhythmia classification based on a combination network of CNN and LSTM Biomed Signal Process Control 2020 57 101819 10.1016/j.bspc.2019.101819 Chen C-Y Automated ECG classification based on 1D deep learning network Methods 2022 202 127 135 10.1016/j.ymeth.2021.04.021 33930574 Cheng X Efficient hardware design of a deep U-net model for pixel-level ECG classification in healthcare device Microelectron J 2022 126 105492 10.1016/j.mejo.2022.105492 Çınar A Tuncer SA Classification of normal sinus rhythm, abnormal arrhythmia and congestive heart failure ECG signals using LSTM and hybrid CNN–SVM deep neural networks Comput Methods Biomech Biomed Engin 2021 24 2 203 214 10.1080/10255842.2020.1821192 32955928 Cortes C Vapnik V Support-vector networks Mach Learn 1995 20 3 273 297 10.1007/BF00994018 Elleuch M Maalej R Kherallah M A new design based-SVM of the CNN classifier architecture with dropout for offline Arabic handwritten recognition Procedia Comput Sci 2016 80 1712 1723 10.1016/j.procs.2016.05.512 Eltrass AS Tayel MB Ammar AI A new automated CNN deep learning approach for identification of ECG congestive heart failure and arrhythmia using constant-Q non-stationary Gabor transform Biomed Signal Process Control 2021 65 102326 10.1016/j.bspc.2020.102326 Eltrass AS Tayel MB Ammar AI Automated ECG multi-class classification system based on combining deep learning features with HRV and ECG measures Neural Comput Appl 2022 34 11 8755 8775 10.1007/s00521-022-06889-z Gaddam PG Sreehari R Automatic classification of cardiac arrhythmias based on ECG signals using transferred deep learning convolution neural network J Phys Conf Ser 2021 2089 1 012058 10.1088/1742-6596/2089/1/012058 Goldberger AL PhysioBank, PhysioToolkit, and PhysioNet: components of a new research resource for complex physiologic signals Circulation 2000 101 23 e215 e220 10.1161/01.CIR.101.23.e215 10851218 Golgowski M, Osowski S (2020) Anomaly detection in ECG using wavelet transformation. In: IEEE 21st International Conference on Computational Problems of Electrical Engineering (CPEE), 2020. IEEE, pp 1–4 Guidotti R Monreale A Ruggieri S Turini F Giannotti F Pedreschi D A survey of methods for explaining black box models ACM Comput Surv (CSUR) 2018 51 5 1 42 10.1145/3236009 Haberl R Jilge G Pulter R Steinbeck G Spectral mapping of the electrocardiogram with Fourier transform for identification of patients with sustained ventricular tachycardia and coronary artery disease Eur Heart J 1989 10 4 316 322 10.1093/oxfordjournals.eurheartj.a059488 2721510 Haykin S Veen BV Signals and Systems 1999 Hoboken Wiley Hua X A novel method for ECG signal classification via one-dimensional convolutional neural network Multimed Syst 2020 1 1 13 Huang J Chen B Yao B He W ECG arrhythmia classification using STFT-based spectrogram and convolutional neural network IEEE Access 2019 7 92871 92880 10.1109/ACCESS.2019.2928017 Iandola FN, Han S, Moskewicz MW, Ashraf K, Dally WJ, Keutzer K (2016) SqueezeNet: AlexNet-level accuracy with 50x fewer parameters and < 0.5 MB model size. arXiv preprint arXiv:1602.07360 Jiao Y Qi H Wu J Capsule network assisted electrocardiogram classification model for smart healthcare Biocybern Biomed Eng 2022 42 2 543 555 10.1016/j.bbe.2022.03.006 Khorrami H Moavenian M A comparative study of DWT, CWT and DCT transformations in ECG arrhythmias classification Expert Syst Appl 2010 37 8 5751 5757 10.1016/j.eswa.2010.02.033 Kiranyaz S Ince T Gabbouj M Real-time patient-specific ECG classification by 1-D convolutional neural networks IEEE Trans Biomed Eng 2015 63 3 664 675 10.1109/TBME.2015.2468589 26285054 Koc M Sut SK Serhatlioglu I Baygin M Tuncer T Automatic prostate cancer detection model based on ensemble VGGNet feature generation and NCA feature selection using magnetic resonance images Multimed Tools Appl 2022 81 5 7125 7144 10.1007/s11042-022-11906-3 Koklu M Ozkan IA Multiclass classification of dry beans using computer vision and machine learning techniques Comput Electron Agric 2020 174 105507 10.1016/j.compag.2020.105507 Krak I, Stelia O, Pashko A, Efremov M, Khorozov O (2020) Electrocardiogram classification using wavelet transformations. In: 2020 IEEE 15th International conference on advanced trends in radioelectronics, telecommunications and computer engineering (TCSET). IEEE, pp 930–933 Krishnakumar S, Yasodha M, Priyadharshini JV, Janney JB, Divakaran S, Christy VL (2021) Detection of arrhythmia and congestive heart failure through classification of ECG signals using deep learning neural network. In: 2021 International conference on advancements in electrical, electronics, communication, computing and automation (ICAECA), 2021. IEEE, pp 1–7 Krizhevsky A Sutskever I Hinton GE Imagenet classification with deep convolutional neural networks Adv Neural Inf Process Syst 2012 25 1097 1105 Kumar A Kumar S Dutt V Dubey AK García-Díaz V IoT-based ECG monitoring for arrhythmia classification using Coyote Grey Wolf optimization-based deep learning CNN classifier Biomed Signal Process Control 2022 76 103638 10.1016/j.bspc.2022.103638 Kumari CU, et al (2020) Heart rhythm abnormality detection and classification using machine learning technique. In: 2020 4th International Conference on Trends in Electronics and Informatics (ICOEI), (48184). IEEE, pp 580–584 Lee HK Choi Y-S Application of continuous wavelet transform and convolutional neural network in decoding motor imagery brain-computer interface Entropy 2019 21 12 1199 10.3390/e21121199 Li C Zheng C Tai C Detection of ECG characteristic points using wavelet transforms IEEE Trans Biomed Eng 1995 42 1 21 28 10.1109/10.362922 7851927 Li J Si Y Xu T Jiang S Deep convolutional neural network based ECG classification system using information fusion and one-hot encoding techniques Math Problems Eng 2018 2018 1 Mahmud T Fattah SA Saquib M Deeparrnet: An efficient deep cnn architecture for automatic arrhythmia detection and classification from denoised ecg beats IEEE Access 2020 8 104788 104800 10.1109/ACCESS.2020.2998788 Meng L Tan W Ma J Wang R Yin X Zhang Y Enhancing dynamic ECG heartbeat classification with lightweight transformer model Artif Intell Med 2022 124 102236 10.1016/j.artmed.2022.102236 35115130 Moody GB Mark RG The impact of the MIT-BIH arrhythmia database IEEE Eng Med Biol Mag 2001 20 3 45 50 10.1109/51.932724 11446209 Nahak S, Saha G (2020) A fusion based classification of normal, arrhythmia and congestive heart failure in ECG. In: 2020 National conference on communications (NCC). IEEE, pp 1–6 Narin A (2020) Detection of focal and non-focal epileptic seizure using continuous wavelet transform-based scalogram images and pre-trained deep neural networks. IRBM Olanrewaju RF Ibrahim SN Asnawi AL Altaf H Classification of ECG signals for detection of arrhythmia and congestive heart failure based on continuous wavelet transform and deep neural networks Indones J Electr Eng Comput Sci 2021 22 3 1520 1528 Ozaltin O Coskun O Yeniay O Subasi A Classification of brain hemorrhage computed tomography images using OzNet hybrid algorithm Int J Imag Syst Technol 2022 2022 1 10.1002/ima.22806 Özaltın Ö, Yeniay Ö (2021) Ecg classification performing feature extraction automatically using a hybrid CNN–SVM algorithm. In: 2021 3rd International congress on human-computer interaction, optimization and robotic applications (HORA). IEEE, pp 1–5 Physionet (2020). https://www.physionet.org/. Accessed 01 June 06 14:00 Pałczyński K Śmigiel S Ledziński D Bujnowski S Study of the few-shot learning for ECG classification based on the PTB-XL dataset Sensors 2022 22 3 904 10.3390/s22030904 35161650 Pławiak P Acharya UR Novel deep genetic ensemble of classifiers for arrhythmia detection using ECG signals Neural Comput Appl 2020 32 15 11137 11161 10.1007/s00521-018-03980-2 Qaisar SM Subasi A Cloud-based ECG monitoring using event-driven ECG acquisition and machine learning techniques Phys Eng Sci Med 2020 43 2 623 634 10.1007/s13246-020-00863-6 32524444 Rahuja N, Valluru SK (2021) A deep neural network approach to automatic multi-class classification of electrocardiogram signals. In: 2021 International Conference on Intelligent Technologies (CONIT). IEEE, pp 1–4 Rajkumar A, Ganesan M, Lavanya R (2019) Arrhythmia classification on ECG using Deep Learning. In: 2019 5th international conference on advanced computing & communication systems (ICACCS). IEEE, pp 365–369 Salem M, Taheri S, Yuan JS (2018) ECG arrhythmia classification using transfer learning from 2-dimensional deep CNN features. In: 2018 IEEE biomedical circuits and systems conference (BioCAS). IEEE, pp 1–4 Sandeep K Kora P Swaraja K Meenakshi K Pampana L ECG classification using machine learning Int J Recent Technol Eng (IJRTE) 2019 8 4 2492 2494 10.35940/ijrte.D6989.118419 Sepahvand M Abdali-Mohammadi F A novel method for reducing arrhythmia classification from 12-lead ECG signals to single-lead ECG with minimal loss of accuracy through teacher-student knowledge distillation Inf Sci 2022 593 64 77 10.1016/j.ins.2022.01.030 Srivastava N Hinton G Krizhevsky A Sutskever I Salakhutdinov R Dropout: a simple way to prevent neural networks from overfitting J Mach Learn Res 2014 15 1 1929 1958 Toma TI, Choi S (2022) A comparative analysis of 2D deep CNN models for arrhythmia detection using STFT-based long duration ECG spectrogram. In: 2022 13th International conference on ubiquitous and future networks (ICUFN). IEEE, pp 483–488 Tuncer T Dogan S Plawiak P Subasi A A novel discrete wavelet-concatenated mesh tree and ternary chess pattern based ECG signal recognition method Biomed Signal Process Control 2022 72 103331 10.1016/j.bspc.2021.103331 Xing Y Accurate ECG classification based on spiking neural network and attentional mechanism for real-time implementation on personal portable devices Electronics 2022 11 12 1889 10.3390/electronics11121889 Xu X Liu H ECG heartbeat classification using convolutional neural networks IEEE Access 2020 8 8614 8619 10.1109/ACCESS.2020.2964749 Zhao Y Cheng J Zhang P Peng X ECG classification using deep CNN improved by wavelet transform Comput Mater Continua 2020 64 3 1615 1628 10.32604/cmc.2020.09938

0

PMC9753894

NO-CC CODE

2022-12-16 23:26:21

no

Soft comput. 2022 Dec 15;:1-20

utf-8

Soft comput

10.1007/s00500-022-07729-x

oa_other

==== Front J Relig Health J Relig Health Journal of Religion and Health 0022-4197 1573-6571 Springer US New York 36520262 1721 10.1007/s10943-022-01721-3 Original Paper (Un)holy Smokes? Religion and Traditional and E-Cigarette Use in the United States http://orcid.org/0000-0003-3798-7753 Hill Terrence D. terrence.hill@utsa.edu 1 Bostean Georgiana gbostean@chapman.edu 2 Upenieks Laura laura_upenieks@baylor.edu 3 Bartkowski John P. john.bartkowski@utsa.edu 4 Ellison Christopher G. christopher.ellison@utsa.edu 4 Burdette Amy M. aburdette@fsu.edu 5 1 grid.215352.2 0000000121845633 Department of Sociology, University of Texas at San Antonio, One UTSA Circle, San Antonio, TX 78249-1644 USA 2 grid.254024.5 0000 0000 9006 1798 Department of Sociology and Environmental Science & Policy Program, Chapman University, Orange, USA 3 grid.252890.4 0000 0001 2111 2894 Department of Sociology, Baylor University, Waco, USA 4 grid.215352.2 0000000121845633 Department of Sociology, University of Texas at San Antonio, San Antonio, USA 5 grid.255986.5 0000 0004 0472 0419 Department of Sociology and Public Health Program, Florida State University, Tallahassee, USA 15 12 2022 126 8 12 2022 © The Author(s), under exclusive licence to Springer Science+Business Media, LLC, part of Springer Nature 2022, Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. This study employed national cross-sectional survey data from the 2021 Crime, Health, and Politics Survey (n = 1578 to 1735) to model traditional cigarette and e-cigarette use as a function of religious affiliation, general religiosity, biblical literalism, religious struggles, and the sense of divine control. Although the odds of abstaining from cigarettes and e-cigarettes were comparable for conservative Protestants and non-affiliates, conservative Protestants were more likely to cut down on cigarettes and e-cigarettes during the pandemic. Religiosity increased the odds of abstaining from cigarettes (not e-cigarettes) and reduced pandemic consumption of cigarettes and e-cigarettes. Biblical literalism was unrelated to abstaining from cigarettes and pandemic changes in cigarette use; however, biblical literalists were more likely to cut e-cigarette use during the pandemic. While the sense of divine control was unrelated to abstaining from cigarettes and e-cigarettes, these beliefs increased the odds of cessation from traditional and e-cigarette use. Finally, our religious struggles index was unrelated to smoking behavior. Our study is among the first to report any association between religion and lower e-cigarette use. Keywords Religion Religiosity Smoking Cigarettes E-cigarettes ==== Body pmcIntroduction Over the past half century, numerous cross-sectional and longitudinal studies have shown that people who are more religious tend to exhibit healthier smoking beliefs and behaviors than their less religious counterparts (Benjamins & Buck, 2008; Clark et al., 1999; Degenhardt et al., 2007; Ford & Hill, 2012; Freeman, 2021; Garrusi & Nakhaee, 2012; Gillum, 2005a, 2005b, 2021; Gottlieb and Green, 1984; Gryczynski & Ward, 2011; Hill et al., 2006; Holt et al., 2015; Idler & Kasl, 1997; Karvinen & Carr, 2014; Kendler et al., 2003; Koenig et al., 1998; Koenig & Vaillant, 2009; Koenig et al., 2012; Mahoney et al., 2005; Nonnemaker et al., 2003, 2006; Parfrey, 1976; Strawbridge et al., 1997, 2001; Stylianou, 2004; Wallace & Forman, 1998; Wang et al., 2015; Ward et al., 2014; Whooley et al., 2002; Yong et al., 2009). Although previous research has emphasized the role of religious attendance, additional protective effects have been observed for religious identities (specific religious groups), private forms of religious behavior (prayer and scripture study), personal orientations and experiences with respect to religion and the divine (intrinsic religiousness, religious salience/importance, positive religious coping, divine relations, and spirituality), and various composite measures of general religiosity. Studies have also examined a wide range of smoking-related outcomes, including negative perceptions of smoking, lifetime smoking, current smoking, former smoking, regular and experimental smoking, smoking initiation, smoking cessation, quitting intentions, number of cigarettes, pack years, clinical nicotine dependence, and objective measures of cotinine (a tobacco alkaloid) in the blood. In perhaps the most comprehensive review of the religion and smoking literature, Koenig et al. (2012) reported that 88% of the 69 highest quality studies reported a protective role of religion. In another review, Garrusi and Nakhaee (2012: 270) concluded that “differences of focus and methodology notwithstanding, most studies have ascertained a deterrent role for religion as regards tobacco use.” Although previous research has made significant contributions to our understanding of religious variations in smoking outcomes, the literature remains surprisingly underdeveloped in several respects. First, the unique contributions of religious affiliation, general religiosity, and specific religious beliefs have been generally understudied and undertheorized. In previous studies, religion measures are often treated as interchangeable with respect to their empirical associations and underlying theoretical explanations. In this context, it is uncommon to see any specific research questions, hypotheses, or theories for specific indicators of religion. Second, little is known about specific religious beliefs, including beliefs concerning scripture (e.g., biblical literalism), God (e.g., the sense of divine control), and religious struggles (e.g., religious doubts). This knowledge gap is limiting because previous theorizing often centers around messages from religious texts and other religious teachings (Garrusi & Nakhaee, 2012; Idler & Kasl, 1997; Mahoney et al., 2005; Strawbridge et al., 2001). Over the past decade, control beliefs have become increasingly important to the broader study of religion and health (Hill et al., 2021a, 2021b; McCullough & Willoughby, 2009; Schieman et al., 2006). There is also mounting evidence of the mental and physical health risks associated with religious struggles (Ellison & Lee, 2010; Hill et al., 2021a, 2021b; Pargament et al., 2001; Upenieks, 2021). Third, only a few studies have considered the association between religion and newer tobacco products like e-cigarettes (Balogh et al., 2018; Hoffmann, 2021; Owotomo & Maslowsky, 2017). This facet of tobacco consumption is important because e-cigarette use is on the rise and has now surpassed traditional cigarette use as the most commonly used tobacco product (Cornelius et al., 2020; Creamer et al., 2019). Although e-cigarettes are often perceived to be more socially acceptable and less physically harmful than traditional cigarettes (Huang et al., 2019; Sæbø & Scheffels, 2017), e-cigarettes contribute to the re-normalization of smoking and to unique public health concerns (Cao et al., 2021; Glantz & Bareham, 2018; Lerner et al., 2015). Glantz and Bareham (2018, p. 215) explain that “while e-cigarettes deliver lower levels of carcinogens than do conventional cigarettes, they still expose users to high levels of ultrafine particles and other toxins that may substantially increase cardiovascular and non-cancer lung disease risks, which account for more than half of all smoking-caused deaths, at rates similar to conventional cigarettes.” Most recently, studies have linked e-cigarette use with an elevated risk of COVID-19 infection (Chen & Kyriakos, 2021; Gaiha et al., 2020; Merianos et al., 2022). Finally, while recent studies of religion and health-related behavior and lifestyles have rightfully concentrated on infectious disease behaviors like mask use, social distancing, and vaccination (Gonzalez et al., 2021; Hill et al., 2020; Perry et al., 2020), researchers have seemingly shifted their focus from traditional chronic disease behaviors like smoking and drinking. This shift is noteworthy because the use of traditional cigarettes remains among the most devastating health behaviors with respect to morbidity and mortality, including the risk of death from COVID-19 (Lariscy et al., 2018; Patanavanich & Glantz, 2021). For these reasons, smoking behavior is considered a lynchpin mechanism of the apparent salutary effects of religious involvement on physical health and mortality (Clark et al., 1999; Gillum et al., 2008; Hill et al., 2017; Hummer et al., 1999; Idler et al., 2017; Koenig & Vaillant, 2009; Strawbridge et al., 1997, 2001). In an effort to build on previous research, we employ national survey data that were collected during the COVID-19 pandemic to formally model the consumption of traditional cigarettes and e-cigarettes as a function of several indicators of religion, including religious affiliation, general religiosity, biblical literalism, religious struggles, and the sense of divine control. In the next section, we summarize relevant research to derive unique hypotheses for each dimension of religion. Because most of the literature is based on traditional cigarette smoking, we have extrapolated many of the ensuing arguments to e-cigarette use a priori. We revisit these assumptions in the discussion section. Background Religious Affiliation Systematic evidence of religious affiliation differences in tobacco use is surprisingly limited. Nevertheless, a few patterns merit brief consideration. First, several studies over the years have reported that persons with no religious affiliation are more prone to smoking cigarettes and using other tobacco products than their counterparts who identify with a specific religious group (Cartwright, 2021, Hussain et al., 2019; Nunziata & Toffolutti, 2019). Second, members of religious groups with clear positions against tobacco use—especially sectarian groups such as the Jehovah’s Witnesses, Seventh-day Adventists, and Latter-day Saints—are especially unlikely to smoke or consume other types of tobacco products (e.g., Jehovah’s Witnesses, 2021; Koenig et al., 2012; Newport, 2013). Beyond these basic patterns, the results are somewhat murky. According to some studies, Protestants (generally), evangelical Protestants (e.g., Baptists, Pentecostals), Black Protestants, and Orthodox Christians are less inclined to smoke than other persons (Cartwright, 2021; Degenhardt et al., 2007; Freeman, 2021; Wasserman & Trovato, 1996). However, other studies deviate from these findings. In their analysis of data collected from a large sample of US adults aged 20 to 32, Whooley et al. (2002) reported that Jews and Presbyterians were among the least likely to smoke, even when compared with evangelicals such as Baptists and Pentecostals. A subsequent national longitudinal study of adolescent smoking behavior showed no protective effects for Baptists, Pentecostals, Catholics, Lutherans, or Methodists (Nonnemaker et al., 2006). Another study of Latinos living in Texas reported that Protestants—nearly all of whom were affiliated with evangelical congregations—were especially unlikely to be current smokers when they attended services regularly, but not when they attended irregularly or not at all (Garcia et al., 2013). Theoretically, religious affiliation should be fundamental to any effects of religion on smoking beliefs and behaviors. Different religious groups are initially socialized to exhibit unique patterns of religiosity, including different norms with respect to public religious activities (religious attendance and participation), private religious activities (prayer, meditation, and scriptural study), and religious salience (the degree to which adherents apply religion to different areas of life). These unique patterns of religiosity contribute to differences in exposure to religion-specific messages concerning sacred texts, the divine, and moral standards for living. Adherents may struggle more or less with the internalization of religious identities and beliefs and the development and maintenance of divine relations. Unique combinations of religiosity, religious beliefs, and struggles can eventually contribute to the ways in which adherents integrate their understanding of divine control into their lives as a framework for meaning-making (e.g., divine attributions of power) and coping (e.g., reliance on the divine for guidance and support). Each of these processes is discussed in greater detail in subsequent sections. Religiosity In contrast to previous studies of religious affiliation, research consistently shows that people who are more religious—indicated by individual measures of religious attendance, prayer frequency, religious salience/importance, and by various composite measures of general religiosity—tend to exhibit healthier smoking beliefs and behaviors (Benjamins & Buck, 2008; Ford & Hill, 2012; Freeman, 2021; Garcia et al., 2013; Gillum, 2005a, 2005b, 2021; Gottlieb & Green, 1984; Gryczynski & Ward, 2011; Hill et al., 2006; Kendler et al., 2003; Koenig et al., 1998; Koenig et al., 2012; Marsiglia et al., 2012; Nonnemaker et al., 2003, 2006; Nunziata & Toffolutti, 2019; Strawbridge et al., 1997, 2001; Stylianou, 2004; Wallace & Forman, 1998; Wallace et al., 2003; Ward et al., 2014; Wasserman & Trovato, 1996; Whooley et al., 2002; Yong et al., 2009). These patterns are impressive in that they have been observed in cross-sectional and longitudinal studies, at various stages of the life course (from adolescence to late life), among women and men, within different racial/ethnic groups (Latinos and non-Hispanic Blacks and Whites), around the world (Australia, Canada, Europe, Malaysia, Mexico, Thailand, and the USA), and across different smoking-related outcomes (negative perceptions of smoking, smoking incidence and prevalence, smoking initiation and cessation, number of cigarettes, pack years, nicotine dependence, and blood-level cotinine). One notable exception to these general patterns is religious media consumption (television and radio). In at least one study of older adults living in North Carolina, religious media consumption was associated with higher rates of current smoking and was unrelated to pack years in the full sample (Koenig et al., 1998). However, when this study’s sample was limited to current smokers, religious media consumption was associated with fewer cigarettes smoked per day. Importantly, religiosity has been consistently unrelated to e-cigarette use in adolescence and young adulthood (Balogh et al., 2018; Hoffmann, 2021; Owotomo & Maslowsky, 2017). Scholars have proposed several ideological, group-based, and psychosocial processes to explain why general religiosity is often associated with healthier smoking outcomes. Ideological explanations suggest that people who are more religious and more engaged with religious institutions have greater exposure to religious teachings that discourage smoking behavior and addiction (Garrusi & Nakhaee, 2012; Gottlieb & Green, 1984; Mahoney et al., 2005; Strawbridge et al., 2001; Whooley et al., 2002). For example, in the Bible, Corinthians (6:19–20) offers the following message: “Or do you not know that your body is a temple of the Holy Spirit within you, which you have from God, and that you are not your own? For you were bought with a price; therefore glorify God in your body.” Although processes related to the sanctification of the body may help to explain lower rates of alcohol consumption and illicit drug use, there is no support for this mechanism in the context of smoking (Mahoney et al., 2005). A related theory points to the internalization of negative attitudes and beliefs concerning the immorality and harmfulness of smoking (Ford & Hill, 2012; Gillum, 2005a; Koenig et al., 1998; Stylianou, 2004; Ward et al., 2014; Yong et al., 2009). In fact, there is direct evidence linking religiosity and lower rates of smoking behaviors through the internalization of anti-smoking sentiments (Ford & Hill, 2012; Ward et al., 2014). Group-based explanations have proposed that people who are more religious and more engaged with religious institutions have greater exposure to non-smoking social networks. Non-smoking reference groups are thought to contribute to implicit norms against smoking behavior (Garrusi & Nakhaee, 2012). There are also more direct mechanisms of social control that are driven by structured time spent with peers and the explicit disapproval of smoking by religious leadership and social network members (Ford & Hill, 2012; Garrusi & Nakhaee, 2012; Gryczynski & Ward, 2011; Nunziata & Toffolutti, 2019; Strawbridge et al., 2001; Ward et al., 2014; Yong et al., 2009). Structured socializing (e.g., regular religious attendance) may limit smoking behavior by increasing exposure to authority figures and by reducing time spent in deviant social networks and routine activities (Garrusi & Nakhaee, 2012; Hoeben et al., 2016). Group-based processes are supported by evidence linking religiosity and healthier smoking behavior through the perceived anti-smoking sentiments of family, peers, and religious leaders (Ford & Hill, 2012; Ward et al., 2014; Yong et al., 2009). Finally, psychosocial explanations suggest that people who are more religious and more engaged with religious institutions are less motivated to smoke because they tend to have more social and psychological resources to manage stress and mental health (Ford & Hill, 2012; Garrusi & Nakhaee, 2012; Gillum, 2005a; Strawbridge et al., 2001; Whooley et al., 2002). The idea is that greater social support (from public religious involvement), the practice of religious coping (feeling supported by a divine other), a general sense of meaning, purpose, and coherence (from organized belief systems and roles in one’s religious group and broader community), and better mental health help to limit the need for smoking as a form of self-medication. For example, Ford and Hill (2012) reported a significant indirect effect of religiosity on any tobacco use in the past year through depressive symptoms. In other words, religiosity contributed to lower rates of tobacco use by reducing depressive symptoms. Scripture Beliefs Although religious doctrine (e.g., body as “temple of the Holy Spirit”) is often invoked to explain religious variations in smoking beliefs and behavior, researchers have yet to formally consider authoritative views of scripture, including widely used measures of biblical literalism or biblical inerrancy. As a basis for our analyses, we nevertheless summarize some indirect and inconsistent evidence from the study of alcohol and drug use in adolescence and young adulthood. In one study of marijuana persistence (used in all three waves), intermittence (used, stopped, used), and desistance (stopped using in one of the final waves) among adolescents and young adults (11–21 years), Ulmer et al. (2010) reported that respondents with stronger beliefs in scriptural inerrancy (The sacred scriptures of your religion are the word of God and are completely without any mistakes.) were less likely to engage in marijuana persistence (versus abstention) and more likely to engage in marijuana persistence (versus desistance). In this same analysis, scriptural inerrancy was unrelated to marijuana initiation (versus abstention), persistence (versus intermittence), abstention (versus desistance), and intermittence (versus desistance). In follow-up studies using these data, this research team showed that scriptural inerrancy was unrelated to marijuana initiation (versus abstention), any marijuana use, and any alcohol use (Desmond et al., 2013; Ulmer et al., 2012). Finally, Koch et al. (2021) recent analysis of college students at 12 universities (including three affiliated with conservative Christian denominations) showed that biblical inerrancy (e.g., The Bible is the infallible word of God.) was associated with lower rates of marijuana use, but not lower rates of alcohol consumption or use of other illicit drugs. If religious doctrine truly helps to explain religious variations in smoking behavior, one would expect more consistent associations between scripture beliefs and substance use. However, available evidence suggests that biblical inerrancy is often unrelated to substance use in adolescence and young adulthood. To our knowledge, there are no previous studies of biblical literalism and smoking behavior in adulthood. While the empirical association between biblical literalism and smoking is uncertain, the idea that the perceived authority of scripture could contribute to variations in smoking beliefs and behavior remains theoretically viable. Religious Struggles Despite growing evidence of the health risks associated with religious struggles (e.g., religious doubts, strained relationships with God, and negative religious coping) (Ellison & Lee, 2010; Hill et al., 2021a, 2021b; Pargament et al., 2001; Upenieks, 2021), little is known about smoking-related outcomes. In fact, our review of the literature revealed only two relevant studies. In one analysis of adult twins from the population-based Virginia Twin Registry, Kendler et al. (2003) showed that stronger beliefs in a judgmental God (e.g., I believe God will punish me if I do something wrong.) were unrelated to nicotine dependence. In the second study, Horton and Loukas (2013) found that negative religious coping (e.g., I feel that stressful situations are God’s way of punishing me for my sins or lack of spirituality.) was also unrelated to the quantity/frequency of cigarettes used in the past month. Although there is no direct empirical evidence linking religious struggles with smoking behavior, the association remains theoretically plausible. Religious struggles refer to “tension and conflict about sacred matters within oneself, with others, and with the supernatural” (Stauner et al., 2016, p. 1). Such tensions and conflicts conceivably challenge the ideological, group-based, and psychosocial processes that would otherwise discourage smoking. Religious doubts could neutralize the moral authority of religious leadership and counter the internalization of religious teachings against smoking. Ideological struggles might also contribute to strained relationships with coreligionists, which could diminish the perceived social costs associated with violating group-based norms against smoking. Finally, the loss of ideological and group-based religious resources could undercut any psychosocial benefits of religious involvement through the loss of meaning and purpose (from ideological uncertainty), supportive social ties (from interpersonal conflicts), and emotional well-being (from the loss of psychosocial resources and the stress of ominous divine relations and beliefs). The Sense of Divine Control The sense of divine control is the belief that “God exerts a commanding authority over the course and direction of one’s life” (Schieman et al., 2006:529). People with a stronger sense of divine control believe that God has decided what their life shall be and depend on God for help and guidance. Although numerous studies have examined the effects of divine control—and related concepts that are both general (God control, involved God, God-mediated control, and locus of God control) and specific (God locus of health control, health God control, and spiritual health locus of control)—on various health-related outcomes (Alyami et al., 2020; Holt et al., 2003; Krause & Rainville, 2022; Krause et al., 2017; Upenieks & Schieman, 2021; Upenieks et al., 2022; Wallston et al., 1999; Welton et al., 1996), little is known about smoking behavior. In fact, we could find only three relevant studies of smoking behavior (Holt et al., 2015; Karvinen & Carr, 2014; Kendler et al., 2003). The first study by Kendler et al. (2003) showed that stronger beliefs in an “involved God” (e.g., God responds to prayers and is very interested in our day-to-day lives) were associated with lower rates of nicotine dependence. Another study, based on a convenience sample of adults, reported no association between God locus of health control and current smoking behavior (Karvinen & Carr, 2014). Finally, Holt et al. (2015) were unable to find any direct or multiplicative effects of active (e.g., Even though I trust that God will take care of me, I still need to take care of myself.) or passive (There is no point in taking care of myself when it’s all up to God anyway.) spiritual health locus of control on regular smoking behavior in a national sample of Black adults. Despite limited empirical evidence with respect to smoking, divine control remains theoretically viable in the sense that these kinds of beliefs may promote or discourage healthier lifestyles or be entirely inconsequential for health behavior. While some studies show that general measures of divine control are associated with healthier behaviors (e.g., lower levels of alcohol consumption) and generally healthy lifestyles (Krause & Rainville, 2022; Welton et al., 1996), others show less healthy behavior (e.g., less exercise) or no associations with specific health behaviors (e.g., diet and sleep quality) and general health lifestyles (Alyami et al., 2020; Krause et al., 2017). Research involving more specific measures of divine health control are similarly mixed, showing healthier behavior, riskier behavior, or no association with health behavior (Alyami et al., 2020; Holt et al., 2003, 2015; Karvinen & Carr, 2014). These inconsistencies could be explained by different styles of divine control beliefs (Holt et al., 2003, 2015; Wallson et al., 1999). People who are more passive in their divine control beliefs (e.g., When good or bad things happen, you see it as part of God’s plan for you.) place more responsibility for their lives and health with God. This orientation may be a form of theological fatalism. By contrast, people who are more active in their divine control beliefs (e.g., You decide what to do without relying on God.) assume more personal responsibility. People who are more collaborative (e.g., All things are possible when I work together with God.) share more responsibility with God. Although there is some evidence to suggest that more passive styles of divine control beliefs are associated with riskier health behavior, the data are far from uniform (Holt et al., 2003, 2015). This body of research suggests a complicated relationship between religion and control beliefs. Although it is intuitive to expect people who are more religious to consistently cede responsibility and control to a higher power, religiosity is associated with greater perceptions of control, including higher levels of the sense of control or mastery, self-control, and health locus of control (Ellison & Burdette, 2012; McCullough & Willoughby, 2009; Pascoe et al., 2016; Schieman, 2008). In fact, the very concept of external attributional style is regularly challenged in the context of religion. Part of this observed pattern is explained by the more active and collaborative dimensions of divine control. Another part is explained by the popular integration of religion and 12-step programs for recovery from addiction (e.g., Alcoholics Anonymous and Narcotics Anonymous). In these contexts, addicts are encouraged to “admit that they are powerless over alcohol, that their lives have become unmanageable” and to “turn their will and lives over to the care of God” (sometimes more generically described as a “Higher Power”) to facilitate a healthier lifestyle (Christo & Franeya, 1995). Hypotheses Informed by the weight of the theoretical and empirical literature, we developed the following five hypotheses to guide our analyses. Hypothesis 1: Conservative Protestants will exhibit higher rates of smoking abstention and cessation than respondents with no religious affiliation. While this group-based difference has the most empirical support, we are much less confident in other group-based comparisons. Hypothesis 2: Respondents who score higher on religiosity will exhibit higher rates of smoking abstention and cessation than other respondents. This hypothesis is generated from the most consistent finding in the religion and smoking literature. Hypothesis 3: Biblical literalists will tend to exhibit higher rates of smoking abstention and cessation than other respondents. This hypothesis is based on the most commonly cited theory for the association between religion and smoking: Religious scripture is an ideological basis for variations in smoking beliefs and behaviors. Hypothesis 4: Respondents who score higher on religious struggles will tend to exhibit lower rates of smoking abstention and cessation than other respondents. Although empirical support for this hypothesis is limited, the theory concerning the various ways in which religious struggles undermine ties to religious institutions is strong. Hypothesis 5: Respondents who score higher on the sense of divine control will tend to exhibit higher rates of smoking abstention and cessation than other respondents. The empirical support for this hypothesis is mixed, but there is enough evidence with respect to smoking and other health-related behaviors to support this expectation. Data To test our hypotheses, we use data from the 2021 Crime, Health, and Politics Survey (CHAPS). The primary purpose of CHAPS is to document the social causes and consequences of various indicators of health and well-being in the USA during the coronavirus (COVID-19) pandemic. CHAPS is based on a national probability sample of 1,771 non-institutionalized adults aged 18 and over living in the USA. Respondents were sampled from the National Opinion Research Center’s (NORC) AmeriSpeak© panel, which is representative of households from all 50 states and the District of Columbia (NORC, 2022). Sampled respondents were invited to complete the online survey in English between May 10, 2021 and June 1, 2021. The data collection process yielded a survey completion rate of 30.7% and a weighted cumulative response rate of 4.4%. The multistage probability sample resulted in a margin of error of ± 3.23% and an average design effect of 1.92. The median self-administered web-based survey lasted approximately 25 min. All respondents were offered the cash equivalent of $8.00 for completing the survey, which is on the more lucrative end of the incentive spectrum for a survey of this duration. The survey was reviewed and approved by the institutional review board at NORC and the review board of the lead author’s university. Informed consent was obtained from all participants. Measures Smoking Abstention and Cessation We measure smoking abstention and cessation with six outcome variables. Respondents were asked the following questions about their regular cigarette use: (a) “Are you a regular smoker of traditional cigarettes, a former smoker, or have you never smoked regularly?” (b) “Are you a regular user of e-cigarettes or smokeless cigarettes, a former user, or have you never used them regularly?” To measure smoking abstention, these items were dummy coded to distinguish (1) respondents who have never smoked regularly and (0) respondents who identify as a regular or former smoker. Our analyses include three abstention outcomes: from all cigarettes, from traditional cigarettes only, and from e-cigarettes only. Respondents were also asked the following questions about recent changes in their cigarette use: (c) “During the coronavirus (COVID-19) pandemic, would you say you have smoked traditional cigarettes more, less, or about the same as before the pandemic?” (d) “During the coronavirus (COVID-19) pandemic, would you say you have used e-cigarettes or smokeless cigarettes more, less, or about the same as before the pandemic?” To measure smoking cessation, these items were dummy coded to distinguish (1) respondents who reported less cigarette use during the pandemic and (0) respondents who reported no change in their behavior or more cigarette use during the pandemic. Our analyses include three cessation outcomes: from all cigarettes, from traditional cigarettes only, and from e-cigarettes only. Religious Affiliation We measure religious affiliation with six dummy variables. These variables capture (a) conservative Protestants (those who reported being Protestant and evangelical/born again), (b) moderate Protestants (those who reported being Protestant without being evangelical/born again), (c) Catholics, (d) other Christians (e.g., those who reported being Mormon, Orthodox, or “just Christian”), (e) other religions (e.g., Jews, Buddhists, and Muslims), and (f) non-affiliates (those with no religious affiliation, including atheists and agnostics). In subsequent analyses, no religious affiliation serves as the common reference group. Religiosity Religious involvement is measured as the mean response to four items. Respondents were asked two questions about their public religious activities: (a) “How often do you usually attend church, synagogue, or other religious meetings?” “How often do you usually attend church, synagogue, or other religious meetings remotely using a computer or phone?” Responses to these questions range from (1) never to (5) several times per week. Respondents were also asked about their private religious activities and the salience of religion in their lives: (c) “How often do you usually spend time in private religious activities such as prayer, meditation, or scriptural study?” (d) “How important is religion in your life today?” Responses to the private activities item range from (1) never to (7) more than once per day. Responses to the importance item range from (1) not important to (5) very important. All items are coded so that higher scores indicate greater religiosity. An exploratory principal components analysis with varimax rotation produced a single component (eigenvalue = 2.77), with loadings ranging from 0.79 to 0.86. A reliability analysis also suggested excellent internal consistency for three items (α = 0.85). Biblical Literalism Biblical literalism is measured with the following item: “Which of these statements comes closest to describing your thoughts about the Bible?” Responses included (1) Bible is true in all ways and should be read literally, word for word. (2) Bible is true in all ways, but should not always be read literally. (3) Bible is mostly true about religious matters, but may contain errors about other things. (4) Bible is not the inspired word of God. This item was dummy coded to distinguish (1) Biblical literalists (response 1) and (0) other respondents (responses 2–4). Religious Struggles Religious struggles are measured as the mean response to four items drawn from the Religious and Spiritual Struggles Scale (Exline et al., 2014). Respondents were asked to indicate how often they (a) “have doubts about their religious or spiritual beliefs,” (b) “feel judged or mistreated by religious or spiritual people,” (c) “feel as though God has abandoned them,” and (d) “feel as though God is punishing them.” Response categories for these items ranged from (1) never to (5) always so that higher index scores would indicate more religious struggles. An exploratory principal components analysis with varimax rotation produced a single component (eigenvalue = 2.28), with loadings ranging from 0.58 to 0.86. A reliability analysis also suggested adequate internal consistency for four items (α = 0.74). The Sense of Divine Control The sense of divine control is measured as the mean response to three items drawn from previous research (Schieman et al., 2005). Respondents were asked the extent to which they agree or disagree with the following statements: (a) “God has decided what my life shall be.” (b) “I decide what to do without relying on God.” (c) “I depend on God for help and guidance.” Response categories for these items ranged from (1) strongly disagree to (5) strongly agree (with reverse coding for item b) so that higher index scores would indicate a greater sense of divine control. An exploratory principal components analysis with varimax rotation produced a single component (eigenvalue = 2.38), with loadings ranging from 0.87 to 0.93. A reliability analysis also suggested excellent internal consistency for three items (α = 0.87). Background Variables Our multivariate analyses include several potential background correlates of our focal variables, including age (continuous years), gender (1 = female; 0 = male), race/ethnicity (dummy variables for Hispanic whites, non-Hispanic black, Latino, and other races/ethnicities), nativity status (1 = US-born; 0 = otherwise), college degree (1 = four-year college degree or higher; 0 = otherwise), employment (1 = employed full- or part-time; 0 = otherwise), annual household income (1 =  < $10,000 to 9 =  ≥ $150,000), financial strain (mean response to three items assessing the extent to which the respondent has trouble paying for health care, monthly bills, and food, α = 0.89), marital status (1 = married; 0 = otherwise), children (1 = presence of child under the age of 18; 0 = otherwise), urbanicity (1 = residence in a large city or town; 0 = otherwise), and region (dummy variables for South, Northeast, Midwest, and West). Analysis Depending on the outcome, our total possible sample size varied from 1755 and 1736 (regular use) to 1606 and 1588 (pandemic use). Due to listwise deletion of missing data, our analytic sample ranged from 1578 to 1735. In other words, over 90% of the total possible sample was retained across regression models. Post-stratification weights were used to assess sampling error and non-response bias. NORC developed post-stratification weights for CHAPS via iterative proportional fitting or raking to general population parameters derived from the Current Population Survey (https://www.census.gov/programs-surveys/cps/data.html). These parameters included age, sex, race/ethnicity, education, and several interactions (age*sex, age*race, and sex*race). Table 1 presents descriptive statistics for all study variables, including variable ranges, sample means, and standard deviations. We then use binary logistic regression to model the odds of lifetime cigarette abstention (Table 2) and cigarette cessation during the pandemic (Table 3).Table 1 Weighted descriptive statistics Range Mean Standard deviation Never Smoked Any Cigarettes 0 to 1 0.58 Never Smoked Trad. Cigarettes 0 to 1 0.59 Never Smoked E-Cigarettes 0 to 1 0.87 Smoked Trad. & E-Cigarettes Less 0 to 1 0.15 Smoked Trad. Cigarettes Less 0 to 1 0.19 Smoked E-Cigarettes Less 0 to 1 0.20 Conservative Protestant 0 to 1 0.21 Moderate Protestant 0 to 1 0.11 Catholic 0 to 1 0.19 Other Christian 0 to 1 0.18 Other Religion 0 to 1 0.05 No Religious Affiliation 0 to 1 0.26 Religiosity − 1.06 to 1.84 − 0.04 0.83 Biblical Literalist 0 to 1 0.20 Religious Struggles 1 to 5 1.92 0.75 The Sense of Divine Control 1 to 5 3.21 1.24 Age 18 to 94 46.90 17.35 Female 0 to 1 0.51 Non-Hispanic White 0 to 1 0.63 Non-Hispanic Black 0 to 1 0.11 Latino 0 to 1 0.16 Other Race/Ethnicity 0 to 1 0.10 US-Born 0 to 1 0.90 College Degree 0 to 1 0.36 Employed 0 to 1 0.60 Household Income 1 to 9 5.51 2.29 Financial Strain 1 to 5 1.73 0.94 Married 0 to 1 0.51 Presence of Child 0 to 1 0.17 Urban Residence 0 to 1 0.29 Southern Resident 0 to 1 0.37 Northeastern Resident 0 to 1 0.17 Midwestern Resident 0 to 1 0.21 Western Resident 0 to 1 0.25 n = 1735 Table 2 Weighted binary logistic regression of never having used cigarettes (abstention) Abstention from traditional and E-Cigarettes Abstention from traditional cigarettes Abstention from E-Cigarettes Conservative Protestant 1.49 (0.97, 2.30) 1.41 (0.91, 2.18) 1.86 (0.92, 3.79) Moderate Protestant 1.09 (0.68, 1.76) 1.12 (0.70, 1.80) 1.42 (0.61, 3.28) Catholic 1.16 (0.75, 1.77) 1.11 (0.72, 1.70) 1.42 (0.68, 2.96) Other Christian 1.08 (0.69, 1.68) 1.11 (0.71, 1.74) 0.96 (0.50, 1.83) Other Religion 0.77 (0.33, 1.76) 0.88 (0.38, 2.07) 0.69 (0.21, 2.23) Religiosity 1.33** (1.07, 1.66) 1.29* (1.04, 1.62) 0.96 (0.69, 1.32) Biblical Literalist 1.17 (0.76, 1.81) 1.29 (0.83, 2.01) 1.62 (0.75, 3.53) Religious Struggles 1.00 (0.82, 1.24) 1.00 (0.81, 1.24) 0.85 (0.65, 1.13) The Sense of Divine Control 0.95 (0.81, 1.12) 0.95 (0.80, 1.12) 0.86 (0.66, 1.12) Shown are unstandardized odds ratios, 95% confidence intervals in parentheses, and two-tailed significance tests: p < 0.05*; p < 0.01**; p < 0.001***. Reference categories include no religious affiliation The estimates for religious affiliation control for age, gender, race/ethnicity, nativity status, education, employment, household income, financial strain, marital status, presence of children, urbanicity, and region The estimates for religiosity, biblical literalism, religious struggles, and the sense of divine control adjust for religious affiliation (no other religion measures) and all background variables n = 1735 Table 3 Weighted binary logistic regression of having smoked fewer cigarettes during the pandemic (cessation) Cessation from traditional and E-Cigarettes Cessation from traditional Cigarettes Cessation from E-Cigarettes Conservative Protestant 3.19** (1.65, 6.20) 1.99* (1.13, 3.49) 2.93*** (1.70, 5.03) Moderate Protestant 1.50 (0.70, 3.12) 1.26 (0.68, 2.34) 1.46 (0.77, 2.77) Catholic 2.74** (1.43, 5.23) 1.76* (1.02, 3.03) 2.24** (1.29, 3.90) Other Christian 2.35* (1.09, 5.10) 1.57 (0.83, 2.96) 1.91* (1.01, 3.63) Other Religion 2.82* (1.02, 7.80) 2.68* (1.10, 6.50) 2.91* (1.16, 7.29) Religiosity 1.57** (1.19, 2.06) 1.57*** (1.22, 2.02) 1.45** (1.12, 1.87) Biblical Literalist 1.59 (0.96, 2.66) 1.47 (0.91, 2.37) 2.40*** (1.49, 3.87) Religious Struggles 0.97 (0.68, 1.38) 0.88 (0.65, 1.19) 1.08 (0.80, 1.46) The Sense of Divine Control 1.25* (1.01, 1.54) 1.25* (1.02, 1.53) 1.27* (1.04, 1.55) Shown are unstandardized odds ratios, 95% confidence intervals in parentheses, and two-tailed significance tests: p < 0.05*; p < 0.01**; p < 0.001***. Reference categories include no religious affiliation The estimates for religious affiliation control for age, gender, race/ethnicity, nativity status, education, employment, household income, financial strain, marital status, presence of children, urbanicity, and region The estimates for religiosity, biblical literalism, religious struggles, and the sense of divine control adjust for religious affiliation (no other religion measures) and all background variables n = 1578 We follow the same analytic strategy in each regression table. The coefficients for religious affiliation control for all background variables. The coefficients for religiosity, biblical literalism, religious struggles, and the sense of divine control adjust for religious affiliation (no other religion measures) and all background variables. All regression models present odds ratios, 95% confidence intervals, and two-tailed statistical tests. Results Descriptive Analyses According to Table 1, the majority of respondents reported never having regularly smoked any cigarettes (58%), traditional cigarettes (59%), and e-cigarettes (87%). Respondents reported smoking both traditional and e-cigarettes (15%), traditional cigarettes (19%), and e-cigarettes (20%) less often during the pandemic. In terms of religious affiliation, the sample included conservative Protestants (21%), moderate Protestants (11%), Catholics (19%), other Christians (18%), respondents of other religious faiths (5%), and respondents with no religious affiliation (26%). The average respondent reported low levels of religiosity and religious struggles and moderate levels of the sense of divine control. Few respondents were classified as biblical literalists (20%). Smoking Abstention In Table 2, we model the odds of never having regularly smoked cigarettes (abstaining). Across outcomes, we failed to observe any differences by religious affiliation, biblical literalism, religious struggles, or the sense of divine control. In other words, the odds of never having smoked regularly were comparable for (a) respondents who identify with a religious group and those who do not, (b) those who believe that the Bible is true in all ways, (c) those who struggle more or less with their religious beliefs and divine relations, and (d) those who believe more or less that God directs and supports their life. Although each unit increase in religiosity raises the odds of abstaining from traditional cigarettes and e-cigarettes (combined) by 33% ([1.33-1] 100) and traditional cigarettes (separately) by 29%, religiosity is unrelated to abstaining from e-cigarette use (separately). Pandemic Smoking Cessation In Table 3, we model the odds of having smoked fewer cigarettes during the pandemic (cessation). Across outcomes, we find several differences by religious affiliation. Compared to respondents with no religious affiliation, conservative Protestants exhibit a 219% increase in the odds of smoking fewer traditional cigarettes and e-cigarettes (combined), a 99% increase in the odds of smoking fewer traditional cigarettes (separately), and a 1.93% increase in the odds of smoking fewer e-cigarettes (separately). We observe substantively identical patterns across outcomes for Catholics and respondents of “other religions.” The odds of smoking fewer traditional cigarettes and e-cigarettes (combined) and e-cigarettes (separately) were higher for “other Christians” than for respondents with no religious affiliation. We failed to observe any differences between moderate Protestants and non-affiliates across outcomes. Religiosity and the sense of divine control are consistently associated with smoking cessation across outcomes. Each unit increase in religiosity raises the odds of smoking fewer traditional cigarettes and e-cigarettes (combined) by 57%, the odds of smoking fewer traditional cigarettes (separately) by 57%, and the odds of smoking fewer e-cigarettes (separately) by 45%. Each unit increase in the sense of divine control raises the odds of smoking fewer traditional cigarettes and e-cigarettes (combined) by 25%, the odds of smoking fewer traditional cigarettes (separately) by 25%, and the odds of smoking fewer e-cigarettes (separately) by 27%. Although biblical literalism is unrelated to cessation from traditional cigarettes and e-cigarettes (combined) and traditional cigarettes (separately), the odds of smoking fewer e-cigarettes (separately) are 140% greater for biblical literalists than for respondents with other views about the Bible. Finally, we failed to observe any differences by religious struggles across outcomes. Discussion Although previous research has made significant contributions to our understanding of religious variations in smoking beliefs and behaviors, unique religion measures have been undertheorized, important religion concepts and new smoking behaviors have been understudied, and important chronic disease behaviors like smoking have been undervalued, particularly during the COVID-19 pandemic. Building on previous work, we employed recently collected national survey data to formally model the consumption of traditional cigarettes and e-cigarettes as a function of several indicators of religion, including religious affiliation, general religiosity, biblical literalism, religious struggles, and the sense of divine control. Our first hypothesis, that conservative Protestants would tend to exhibit higher rates of smoking abstention and cessation than respondents with no religious affiliation, received mixed support. On the one hand, the odds of abstaining from traditional cigarettes and e-cigarettes were comparable for conservative Protestants and respondents with no religious affiliation. Although these findings are inconsistent with some studies (Freeman, 2021; Garcia et al., 2013; Wasserman & Trovato, 1996), they are not without precedent (Degenhardt et al., 2007; Nonnemaker et al., 2006). On the other hand, conservative Protestants (and Catholics) were more likely than non-affiliates to have cut their consumption of traditional cigarettes and e-cigarettes during the pandemic. Our study is the first of which we are aware to document an association between religious affiliation and e-cigarette use. Our second hypothesis was that respondents who score higher on religiosity would tend to exhibit higher rates of smoking abstention and cessation than other respondents. This result was our most consistent finding. Religiosity increased the odds abstaining from traditional cigarettes and increased the odds of cessation from traditional cigarettes and e-cigarettes during the pandemic. Although our findings for traditional cigarette use support numerous studies (see Garrusi & Nakhaee, 2012; Koenig et al., 2012 for reviews), we are the first to document an association between religiosity and e-cigarette use. We were unable to observe any association between religiosity and abstaining from e-cigarettes, which is also consistent with previous research (Balogh et al., 2018; Hoffmann, 2021; Owotomo & Maslowsky, 2017). Our third hypothesis was that biblical literalists would tend to exhibit higher rates of smoking abstention and cessation than other respondents. We found little support for this hypothesis. Biblical literalism was unrelated to abstaining from traditional cigarettes and e-cigarettes and pandemic changes in traditional cigarette use. These results are generally consistent with previous studies showing no association between scriptural inerrancy and the use of marijuana and alcohol in adolescence and young adulthood (Desmond et al., 2013; Ulmer et al., 2012). We found some evidence to suggest that biblical literalists were more likely to have cut e-cigarette use during the pandemic. This pattern is consistent with previous studies of scriptural inerrancy and marijuana use (Koch et al. 2021; Ulmer et al., 2010). Our fourth hypothesis stated that respondents who score higher on religious struggles would tend to exhibit lower rates of smoking abstention and cessation than other respondents. We found no support for this hypothesis in our main analyses. In fact, our religious struggles index was unrelated to all four of our smoking outcomes. These null patterns confirm previous work with data collected from a national sample of adults and multiple smoking outcomes (Horton & Loukas, 2013; Kendler et al., 2003). We note that our supplemental analyses indicated that two indicators of religious struggles (feeling abandoned by God and judged or mistreated by religious or spiritual people) reduced the odds of abstaining from e-cigarettes. These patterns, which are unprecedented in the literature, are consistent with our fourth hypothesis. Our fifth and final hypothesis suggested that respondents who score higher on the sense of divine control would tend to exhibit higher rates of smoking abstention and cessation than other respondents. Again, we found mixed support for this hypothesis. While the sense of divine control was unrelated to abstaining from traditional and e-cigarettes (combined and separately), these beliefs increased the odds of cessation from traditional and e-cigarettes (combined and separately). The null patterns for the sense of divine control support some previous studies (Holt et al., 2015; Karvinen & Carr, 2014). However, the finding that divine control could increase the odds of cessation is more consistent with previous studies linking divine control beliefs with healthier behaviors like lower levels of nicotine dependence and alcohol consumption (Holt et al., 2015; Kendler et al., 2003; Krause & Rainville, 2022; Welton et al., 1996). The inconsistencies in our findings could be due to the fact that our measure of divine control is weighted more toward the distribution of power (possibly leading to personal quiescence or fatalism) than toward a perceived collaboration with God. While previous studies tend to focus on one or two seemingly interchangeable religion measures and traditional cigarette use, we contribute to previous work by examining the unique effects of religious affiliation, religiosity, biblical literalism, religious struggles, and the sense of divine control on traditional cigarette use, e-cigarette use, and pandemic smoking behavior. Although religion was unrelated to lifetime abstention from e-cigarettes, several indicators of religion (religious affiliation, religiosity, biblical literalism, and the sense of divine control) increased the odds of cessation from e-cigarettes during the pandemic. It is likely that few institutions of religion have directly addressed the morality of e-cigarette use. This makes sense because e-cigarettes have only recently emerged in the USA over the last two decades. Prior to the FDA’s “deeming rule” in 2016, many e-cigarettes were marketed as cessation devices. While many people perceive e-cigarettes to be safer than cigarettes, the messaging around the health effects of traditional cigarettes have been unambivalent and in place for a generation. These differences in norms and perceptions of safety and acceptability of traditional cigarettes versus e-cigarettes may contribute to the null associations with lifetime abstention. We note that similar patterns have been observed for other “morally ambiguous” outcomes like prescription drugs and medical marijuana (Burdette et al. 2018a, 2018b). This combination of results suggests a recent activation of the deterrent role of religion with respect to e-cigarette use during the pandemic (i.e., no effects for lifetime abstention, but consistent effects for pandemic cessation). While this finding is difficult to interpret, two potential explanations deserve consideration. First, it is possible that religious networks are particularly well suited to transmitting relatively new information about the potential harms of somewhat novel substances such as e-cigarettes. A good deal of research has focused on the transposability (transference, portability) of religious schemas (e.g., Shah et al., 2016). It is possible that religious schemas (durable interpretive frameworks) that have long conveyed opposition to traditional cigarette use are readily parlayed into concerns and cautions about e-cigarettes among people of faith. Second, it is quite likely that some religious groups and people understood the pandemic as a possible “end times” cataclysm or at least a morally significant event (even, for some, a sign from God). Their religious viewpoint may have prompted them to engage in health behavior changes that amounted to “cleaning up” their body-as-temple, at least where e-cigarettes were concerned. Clearly, additional research is needed to determine if this pattern is observed concerning other substances. And the use of other methods (e.g., qualitative inquiry) is needed to explore these possibilities in greater detail. In our analyses, religiosity stands out as the only consistent predictor of smoking behavior. This finding is notable because the individual items in the religiosity index are inconsistently associated with smoking behavior. While some items are unrelated to smoking (e.g., religious importance), others are positively associated with smoking (e.g., prayer). Prayer may be a reactive response, even a petition for help, to intended cessation that is difficult to implement and sustain. In the end, physical attendance (not virtual attendance) and the synergistic effects of the combined religiosity index are the most reliable predictors of smoking abstinence and cessation. In contrast to general religiosity, the protective effects of religious affiliation and biblical literalism were more sporadic. Given that the protective effects of religious affiliation were entirely explained by more robust levels of general religiosity (not shown), there were no residual ideological effects net of adjustments for public and private religious activities. The effects of religious struggles and the sense of divine control also range from non-existent to insalubrious. These patterns, along with the more pronounced pandemic effects, lead us to prioritize group-based and psychosocial processes over more enduring and stable religion-specific ideological mechanisms when explaining variations in smoking behavior. Study Limitations We note that our analyses are limited by our cross-sectional design and self-reported data. Although we have emphasized theoretical explanations that imply a true causal association between religiosity and smoking-related behavior, our analyses are vulnerable to two alternative or artifactual explanations. The first alternative explanation suggests that the apparent protective effects of religious involvement on smoking are due to social desirability (Gillum, 2005a). In other words, people who are more religious and more engaged with religious institutions may be motivated to lie about their smoking beliefs and behaviors to present a consistent religious identity. This possibility has been addressed and dismissed by studies linking religiosity (religious attendance) with lower levels of cotinine in the blood (Gillum, 2005a, 2005b, 2021). The logic is that nicotine biomarkers are not subject to the same social desirability processes that threaten self-reports of smoking. If religious people were systematically underreporting their smoking behavior, religiosity would be unrelated to more objective assessments of smoking behavior. These data are also relevant to group-based explanations. The level of cotinine in the blood is a general indicator of exposure to tobacco smoke from all sources, including second-hand environmental exposures. This research is consistent with the notions of more religious people being more socially integrated into non-smoking social networks and more personally motivated and equipped to avoid smoking. The second alternative explanation suggests that the apparent protective effects of religious involvement on smoking are due to health behavior selection (Whooley et al., 2002). In other words, people who engage in behaviors that are normative to a group are more likely to become members or remain as members of the group. Alternatively, if people engage in behaviors that are non-normative or stigmatized by the group, they may be socially disqualified or rejected from the group. This concern is addressed by research showing an inverse association between religiosity and smoking when controlling for prior smoking status and when the sample is limited to smokers (Koenig et al., 1998). In these designs, the protective effects of public and private forms of religiosity are observed when smoker status is adjusted or held constant. Of course, this second alternative is also lent force by considerations of religion and health as “structuring structures,” a concept first popularized by Bourdieu (1990). From this vantage point, lifestyle practices in the social fields of religion and health may be linked through mutual reinforcement with complementary logics (wellness advocacy in the health field, bodily sanctification in the religious field). This perspective can help researchers to delve more deeply into what are sometimes called (or criticized as) “selection effects.” Yet, these observed forms of “selectivity” are, in fact, a product of living in complex, cascading social worlds that sometimes overlap and thereby foster attitudinal or behavioral reinforcement. More theoretically grounded research with rich data is needed to explore this process in greater detail. Conclusion Although we are confident that religiosity and, to a lesser extent, religious affiliation and biblical literalism can play a deterrent role with respect to smoking behavior, additional research is needed to replicate our findings with longitudinal data, more objective measures of smoking, and more active and collaborative assessments of divine control (e.g., active spiritual health locus of control and God-mediated control). In this study, we emphasized the combined effects of general religiosity, but future research will need to unpack the unique role of prayer, including different types of prayer. Additional studies assessing mediation are needed to establish any ideological, group-based, or psychosocial mechanisms of religious variations in smoking behavior. Finally, we must also take stock of the moderating role of religion. For example, the sense of divine control could moderate or buffer the effects of known risk factors for smoking (e.g., pandemic stress). These are just a few directions to improve our understanding of religion and smoking behavior. Finally, while we have no clinical expertise as social scientists, there may be some practical benefits associated with providing an option for the delivery of tobacco cessation services in cooperation with faith-based organizations for clients who prefer the integration of spiritual or religious elements into such interventions. Of course, any such partnerships should proceed with great caution to ensure that client choice in program content is fully respected, and we leave the implementation of such ventures to those with greater expertise in the health promotion field. Funding None. Declarations Conflict of interest All authors declare no conflict of interest. Informed consent Because this article employed secondary data that were previously collected de-identified data to protect respondents, it was exempt from human subjects review. When the study was collected, informed consent was obtained from all individual participants. No animals were included in the study. Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References Alyami M Serlachius A Mokhtar I Broadbent E The association of illness perceptions and God locus of health control with self-care behaviours in patients with type 2 diabetes in Saudi Arabia Health Psychology and Behavioral Medicine 2020 8 329 348 10.1080/21642850.2020.1805322 34040875 Balogh E Faubl N Riemenschneider H Balázs P Bergmann A Cseh K Horváth F Schelling J Terebessy A Kiss I Cigarette, waterpipe and e-cigarette use among an international sample of medical students. Cross-sectional multicenter study in Germany and Hungary BMC Public Health 2018 18 91 10.1186/s12889-018-5494-6 Benjamins MR Buck AC Religion: A sociocultural predictor of health behaviors in Mexico Journal of Aging and Health 2008 20 290 305 10.1177/0898264308315429 18263855 Bourdieu P The logic of practice 1990 Polity Press Burdette AM Hill TD Webb NS Ford JA Haynes SH Religious involvement and substance use among urban mothers Journal for the Scientific Study of Religion 2018 57 156 172 10.1111/jssr.12501 Burdette AM Webb NS Hill TD Haynes SH Ford JA Religious involvement and marijuana use for medical and recreational purposes Journal of Drug Issues 2018 48 421 434 10.1177/0022042618770393 29899577 Cao Y Wu D Ma Y Ma X Wang S Li F Li M Zhang T Toxicity of electronic cigarettes: A general review of the origins, health hazards, and toxicity mechanisms Science of the Total Environment 2021 772 145475 10.1016/j.scitotenv.2021.145475 33770885 Cartwright, K. (2021). Delivered from the temptation of smoking: An examination of religion and health behaviors of new US immigrants. Journal of Religion and Health, 60, 1739–1759. 10.1007/s10943-019-00843-5 Chen DTH Kyriakos CN Cigarette and E-cigarettes dual users, exclusive users and COVID-19: Findings from four UK birth cohort studies International Journal of Environmental Research and Public Health 2021 18 3935 10.3390/ijerph18083935 33918098 Christo G Franey C Drug users' spiritual beliefs, locus of control and the disease concept in relation to Narcotics Anonymous attendance and six-month outcomes Drug and Alcohol Dependence 1995 38 51 56 10.1016/0376-8716(95)01103-6 7648997 Clark KM Friedman HS Martin LR A longitudinal study of religiosity and mortality risk Journal of Health Psychology 1999 4 381 391 10.1177/135910539900400307 22021605 Cornelius ME Wang TW Jamal A Loretan CG Neff LJ Tobacco product use among adults—United States, 2019 Morbidity and Mortality Weekly Report 2020 69 1736 1742 10.15585/mmwr.mm6946a4 33211681 Creamer MR Wang TW Babb S Cullen KA Day H Willis G Jamal A Neff L Tobacco product use and cessation indicators among adults—United States, 2018 Morbidity and Mortality Weekly Report 2019 68 1013 1019 10.15585/mmwr.mm6845a2 Degenhardt L Chiu WT Sampson N Kessler RC Anthony JC Epidemiological patterns of extra-medical drug use in the United States: Evidence from the National Comorbidity Survey Replication, 2001–2003 Drug and Alcohol Dependence 2007 90 210 223 10.1016/j.drugalcdep.2007.03.007 17481828 Desmond SA Ulmer JT Bader CD Religion, self control, and substance use Deviant Behavior 2013 34 384 406 10.1080/01639625.2012.726170 Ellison CG Burdette AM Religion and the sense of control among US adults Sociology of Religion 2012 73 1 22 10.1093/socrel/srr035 Ellison CG Lee J Spiritual struggles and psychological distress: Is there a dark side of religion? Social Indicators Research 2010 98 501 517 10.1007/s11205-009-9553-3 Exline JJ Pargament KI Grubbs JB Yali AM The Religious and Spiritual Struggles Scale: Development and initial validation Psychology of Religion and Spirituality 2014 6 208 222 10.1037/a0036465 Ford JA Hill TD Religiosity and adolescent substance use: Evidence from the national survey on drug use and health Substance Use & Misuse 2012 47 787 798 10.3109/10826084.2012.667489 22443107 Freeman JA Does the Association between Religiosity and Substance Use Vary between Adolescence and Early Adulthood? Sociological Inquiry 2021 91 3 29 10.1111/soin.12331 Gaiha SM Cheng J Halpern-Felsher B Association between youth smoking, electronic cigarette use, and COVID-19 Journal of Adolescent Health 2020 67 519 523 10.1016/j.jadohealth.2020.07.002 Garcia G Ellison CG Sunil TS Hill TD Religion and selected health behaviors among Latinos in Texas Journal of Religion and Health 2013 52 18 31 10.1007/s10943-012-9640-7 22911394 Garrusi B Nakhaee N Religion and smoking: A review of recent literature The International Journal of Psychiatry in Medicine 2012 43 279 292 10.2190/PM.43.3.g 22978085 Gillum RF Frequency of attendance at religious services and cigarette smoking in American women and men: The Third National Health and Nutrition Examination Survey Preventive Medicine 2005 41 607 613 10.1016/j.ypmed.2004.12.006 15917059 Gillum RF Religiosity and the validity of self-reported smoking: The third national health and nutrition examination survey Review of Religious Research 2005 47 190 196 10.2307/3512050 Gillum RF Frequency of attendance at religious services and exposure to environmental tobacco smoke Journal of Religion and Health 2021 60 1760 1765 10.1007/s10943-020-01130-4 33389437 Gillum RF King DE Obisesan TO Koenig HG Frequency of attendance at religious services and mortality in a US national cohort Annals of Epidemiology 2008 18 124 129 10.1016/j.annepidem.2007.10.015 18083539 Glantz SA Bareham DW E-cigarettes: Use, effects on smoking, risks, and policy implications Annual Review of Public Health 2018 39 215 235 10.1146/annurev-publhealth-040617-013757 Gonzalez KE James R Bjorklund ET Hill TD Conservatism and infrequent mask usage: A study of US counties during the novel coronavirus (COVID-19) pandemic Social Science Quarterly 2021 102 2368 2382 10.1111/ssqu.13025 34908612 Gottlieb NH Green LW Life events, social network, life-style, and health: An analysis of the 1979 national survey of personal health practices and consequences Health Education Quarterly 1984 11 91 105 10.1177/1090198184011001 6511433 Gryczynski J Ward BW Social norms and the relationship between cigarette use and religiosity among adolescents in the United States Health Education & Behavior 2011 38 39 48 10.1177/1090198110372331 21189421 Hill TD Burdette AM Ellison CG Musick MA Religious attendance and the health behaviors of Texas adults Preventive Medicine 2006 42 309 312 10.1016/j.ypmed.2005.12.005 16445971 Hill TD Gonzalez K Burdette AM The blood of Christ compels them: State religiosity and state population mobility during the coronavirus (COVID-19) pandemic Journal of Religion and Health 2020 59 2229 2242 10.1007/s10943-020-01058-9 32596753 Hill T Upenieks L Ellison C Angel J López Ortega M Miguel Gutierrez Robledo L Religious involvement, health locus of control, and sleep disturbance: A study of older Mexican Americans Understanding the context of cognitive aging: Mexico and the United States 2021 Cham, Switzerland Springer Nature 161 179 Hill TD Vaghela P Ellison CG Rote S Processes linking religious involvement and telomere length Biodemography and Social Biology 2017 63 167 188 10.1080/19485565.2017.1311204 28521619 Hill TD Zeng L Rambotti S Mossakowski KN Johnson RJ Sad eyes, crooked crosses: Religious struggles, Psychological distress and the mediating role of psychosocial resources Journal of Religion and Health 2021 60 2573 2591 10.1007/s10943-021-01273-y 33970412 Hoeben EM Meldrum RC Walker DA Young JT The role of peer delinquency and unstructured socializing in explaining delinquency and substance use: A state-of-the-art review Journal of Criminal Justice 2016 47 108 122 10.1016/j.jcrimjus.2016.08.001 Hoffmann JP Sensation seeking and adolescent e-cigarette use Journal of Substance Use 2021 26 542 548 10.1080/14659891.2020.1856209 Holt CL Clark EM Kreuter MW Rubio DM Spiritual health locus of control and breast cancer beliefs among urban African American women Health Psychology 2003 22 294 299 10.1037/0278-6133.22.3.294 12790257 Holt CL Roth DL Huang J Clark EM Gender differences in the roles of religion and locus of control on alcohol use and smoking among African Americans Journal of Studies on Alcohol and Drugs 2015 76 482 492 10.15288/jsad.2015.76.482 25978836 Horton KD Loukas A Depressive symptoms, religious coping, and cigarette smoking among post-secondary vocational students Psychology of Addictive Behaviors 2013 27 705 713 10.1037/a0031195 23276324 Huang J Feng B Weaver SR Pechacek TF Slovic P Eriksen MP Changing perceptions of harm of e-cigarette vs cigarette use among adults in 2 US national surveys from 2012 to 2017 JAMA Network Open 2019 2 e191047 10.1001/jamanetworkopen.2019.1047 30924893 Hummer RA Rogers RG Nam CB Ellison CG Religious involvement and US adult mortality Demography 1999 36 273 285 10.2307/2648114 10332617 Hussain M Walker C Moon G Smoking and religion: Untangling associations using English survey data Journal of Religion and Health 2019 58 2263 2276 10.1007/s10943-017-0434-9 28667475 Idler E Blevins J Kiser M Hogue C Religion, a social determinant of mortality? A 10-year follow-up of the Health and Retirement Study PLoS ONE 2017 12 e0189134 10.1371/journal.pone.0189134 29261682 Idler EL Kasl SV Religion among disabled and nondisabled persons I: Cross-sectional patterns in health practices, social activities, and well-being Ellen The Journals of Gerontology Series B: Psychological Sciences and Social Sciences 1997 52 S294 S305 10.1093/geronb/52B.6.S294 9403523 Jehovah’s Witnesses. (2021). Is smoking a sin? The Bible’s answer. https://www.jw.org/en/bible-teachings/questions/smoking-sindownloaded 12/21/2021 Karvinen KH Carr LJ Does the perception that God controls health outcomes matter for health behaviors? Journal of Health Psychology 2014 19 521 530 10.1177/1359105312474914 23431129 Kendler KS Liu XQ Gardner CO McCullough ME Larson D Prescott CA Dimensions of religiosity and their relationship to lifetime psychiatric and substance use disorders American Journal of Psychiatry 2003 160 496 503 10.1176/appi.ajp.160.3.496 12611831 Koch JR Wagner BG Roberts AE Christian universities as moral communities: Drinking, sex, and drug use among university students in the United States The Social Science Journal 2021 10.1080/03623319.2021.1963108 Koenig HG George LK Cohen HJ Hays JC Larson DB Blazer DG The relationship between religious activities and cigarette smoking in older adults The Journals of Gerontology Series a: Biological Sciences and Medical Sciences 1998 53 M426 M434 10.1093/gerona/53A.6.M426 9823746 Koenig H Koenig HG King D Carson VB Handbook of religion and health 2012 Oxford University Press Koenig LB Vaillant GE A prospective study of church attendance and health over the lifespan Health Psychology 2009 28 117 124 10.1037/a0012984 19210025 Krause N Hill PC Emmons R Pargament KI Ironson G Assessing the relationship between religious involvement and health behaviors Health Education & Behavior 2017 44 278 284 10.1177/1090198116655314 27387205 Krause N Rainville G Participation in combat, god-mediated control beliefs, and alcohol consumption Mental Health, Religion &amp; Culture 2022 25 320 331 10.1080/13674676.2021.2005009 Lariscy JT Hummer RA Rogers RG Cigarette smoking and all-cause and cause-specific adult mortality in the United States Demography 2018 55 1855 1885 10.1007/s13524-018-0707-2 30232778 Lerner CA Sundar IK Watson RM Elder A Jones R Done D Kurtzman R Ossip DJ Robinson R McIntosh S Rahman I Environmental health hazards of e-cigarettes and their components: Oxidants and copper in e-cigarette aerosols Environmental Pollution 2015 198 100 107 10.1016/j.envpol.2014.12.033 25577651 Mahoney A Carels RA Pargament KI Wachholtz A Edwards Leeper L Kaplar M Frutchey R The sanctification of the body and behavioral health patterns of college students The International Journal for the Psychology of Religion 2005 15 221 238 10.1207/s15327582ijpr1503_3 Marsiglia FF Ayers SL Hoffman S Religiosity and adolescent substance use in Central Mexico: Exploring the influence of internal and external religiosity on cigarette and alcohol use American Journal of Community Psychology 2012 49 87 97 10.1007/s10464-011-9439-9 21533659 McCullough ME Willoughby BL Religion, self-regulation, and self-control: Associations, explanations, and implications Psychological Bulletin 2009 135 69 93 10.1037/a0014213 19210054 Merianos AL Russell AM Mahabee-Gittens EM Barry AE Yang M Lin HC Concurrent use of e-cigarettes and cannabis and associated COVID-19 symptoms, testing, and diagnosis among student e-cigarette users at four US Universities Addictive Behaviors 2022 126 107170 10.1016/j.addbeh.2021.107170 34776303 Newport, F. (2013). In U.S., strong link between church attendance, smoking. https://news.gallup.com/poll/163856/strong-link-church-attendance-smoking.aspx?version=printdownloaded 4/27/2022. Nonnemaker JM McNeely CA Blum RW Public and private domains of religiosity and adolescent health risk behaviors: Evidence from the National Longitudinal Study of Adolescent Health Social Science & Medicine 2003 57 2049 2054 10.1016/S0277-9536(03)00096-0 14512236 Nonnemaker J McNeely CA Blum RW Public and private domains of religiosity and adolescent smoking transitions Social Science & Medicine 2006 62 3084 3095 10.1016/j.socscimed.2005.11.052 16423435 NORC. (2022). Technical Overview of the AMERISPEAK® Panel NORC's Probability-Based Household Panel. National Opinion Research Center (NORC), University of Chicago, USA [Update: February 2022]. https://amerispeak.norc.org/content/dam/amerispeak/research/pdf/AmeriSpeak%20Technical%20Overview%202019%2002%2018.pdf Nunziata L Toffolutti V “Thou shalt not smoke”: Religion and smoking in a natural experiment of History SSM-Population Health 2019 8 100412 10.1016/j.ssmph.2019.100412 31338409 Owotomo O Maslowsky J Marijuana and e-cigarette use in a US national sample of 8th and 10th grade never-smokers of conventional cigarettes Journal of Applied Research on Children: Informing Policy for Children at Risk 2017 8 1 17 Parfrey PS The effect of religious factors on intoxicant use Scandinavian Journal of Social Medicine 1976 4 135 140 10.1177/140349487600400324 996473 Pargament KI Koenig HG Tarakeshwar N Hahn J Religious struggle as a predictor of mortality among medically ill elderly patients: A 2-year longitudinal study Archives of Internal Medicine 2001 161 1881 1885 10.1001/archinte.161.15.1881 11493130 Pascoe AE Hill TD Mossakowski KN Johnson RJ Religious involvement and perceptions of control: Evidence from the Miami-Dade Health Survey Journal of Religion and Health 2016 55 862 873 10.1007/s10943-015-0081-y 26118384 Patanavanich R Glantz SA Smoking is associated with worse outcomes of COVID-19 particularly among younger adults: A systematic review and meta-analysis BMC Public Health 2021 21 1 9 10.1186/s12889-021-11579-x 33388037 Perry SL Whitehead AL Grubbs JB Culture wars and COVID-19 conduct: Christian nationalism, religiosity, and Americans’ behavior during the coronavirus pandemic Journal for the Scientific Study of Religion 2020 59 405 416 10.1111/jssr.12677 Sæbø G Scheffels J Assessing notions of denormalization and renormalization of smoking in light of e-cigarette regulation International Journal of Drug Policy 2017 49 58 64 10.1016/j.drugpo.2017.07.026 28987929 Schieman S The religious role and the sense of personal control Sociology of Religion 2008 69 273 296 10.1093/socrel/69.3.273 Schieman S Pudrovska T Milkie MA The sense of divine control and the self-concept: A study of race differences in late life Research on Aging 2005 27 165 196 10.1177/0164027504270489 Schieman S Pudrovska T Pearlin LI Ellison CG The sense of divine control and psychological distress: Variations across race and socioeconomic status Journal for the Scientific Study of Religion 2006 45 529 549 10.1111/j.1468-5906.2006.00326.x Shah S Bartkowski JP Xu X Gendered God imagery and attitudes toward mothers’ labor force participation: Examining the transposable character of religious schemas Journal for the Scientific Study of Religion 2016 55 540 557 10.1111/jssr.12279 Stauner N Exline JJ Grubbs JB Pargament KI Bradley DF Uzdavines A Bifactor models of religious and spiritual struggles: Distinct from religiousness and distress Religions 2016 7 68 10.3390/rel7060068 Strawbridge WJ Cohen RD Shema SJ Kaplan GA Frequent attendance at religious services and mortality over 28 years American Journal of Public Health 1997 87 957 961 10.2105/AJPH.87.6.957 9224176 Strawbridge WJ Shema SJ Cohen RD Kaplan GA Religious attendance increases survival by improving and maintaining good health behaviors, mental health, and social relationships Annals of Behavioral Medicine 2001 23 68 74 10.1207/S15324796ABM2301_10 11302358 Stylianou S The role of religiosity in the opposition to drug use International Journal of Offender Therapy and Comparative Criminology 2004 48 429 448 10.1207/S15324796ABM2301_10 15245655 Ulmer JT Desmond SA Jang SJ Johnson BR Teenage religiosity and changes in marijuana use during the transition to adulthood Interdisciplinary Journal of Research on Religion 2010 6 1 19 Ulmer JT Desmond SA Jang SJ Johnson BR Religious involvement and dynamics of marijuana use: Initiation, persistence, and desistence Deviant Behavior 2012 33 448 468 10.1080/01639625.2011.636653 Upenieks L Changes in religious doubt and physical and mental health in emerging adulthood Journal for the Scientific Study of Religion 2021 60 332 361 10.1111/jssr.12712 Upenieks L Schieman S The belief in divine control and the mental health effects of stressful life events: A study of education-based contingencies Review of Religious Research 2021 63 183 215 10.1007/s13644-020-00426-x Upenieks L Schieman S Bierman A Jitters on the eve of the great recession: Is the belief in divine control a protective resource? Sociology of Religion 2022 83 194 221 10.1093/socrel/srab018 Wallace JM Jr Brown TN Bachman JG Laveist TA The influence of race and religion on abstinence from alcohol, cigarettes and marijuana among adolescents Journal of Studies on Alcohol 2003 64 843 848 10.15288/jsa.2003.64.843 14743948 Wallace JM Jr Forman TA Religion's role in promoting health and reducing risk among American youth Health Education & Behavior 1998 25 721 741 10.1177/109019819802500604 9813744 Wallston KA Malcarne VL Flores L Hansdottir I Smith CA Stein MJ Weisman MH Clements PJ Does God determine your health? The God locus of health control scale Cognitive Therapy and Research 1999 23 2 131 142 10.1023/A:1018723010685 Wang Z Koenig HG Al Shohaib S Religious involvement and tobacco use in mainland China: A preliminary study BMC Public Health 2015 15 155 10.1186/s12889-015-1478-y 25886594 Ward BW Allen A Gryczynski J Racial/ethnic differences in the relationship among cigarette use, religiosity, and social norms for US adolescents Journal of Ethnicity in Substance Abuse 2014 13 337 361 10.1080/15332640.2014.958636 25397636 Wasserman I Trovato F The influence of religion on smoking and alcohol consumption: Alberta case study International Review of Modern Sociology 1996 26 43 56 Welton GL Adkins AG Ingle SL Dixon WA God control: The fourth dimension Journal of Psychology and Theology 1996 24 13 25 10.1177/009164719602400102 Whooley MA Boyd AL Gardin JM Williams DR Religious involvement and cigarette smoking in young adults: The CARDIA study Archives of Internal Medicine 2002 162 1604 1610 10.1001/archinte.162.14.1604 12123404 Yong HH Hamann SL Borland R Fong GT Omar M Adult smokers' perception of the role of religion and religious leadership on smoking and association with quitting: A comparison between Thai Buddhists and Malaysian Muslims Social Science & Medicine 2009 69 1025 1031 10.1016/j.socscimed.2009.07.042 19695758

36520262

PMC9753896

NO-CC CODE

2022-12-16 23:26:35

no

J Relig Health. 2022 Dec 15;:1-26

utf-8

J Relig Health

10.1007/s10943-022-01721-3

oa_other

==== Front Lancet Lancet Lancet (London, England) 0140-6736 1474-547X Elsevier Ltd. S0140-6736(21)00769-8 10.1016/S0140-6736(21)00769-8 Editorial Racism in the USA: ensuring Asian American health equity The Lancet 1 4 2021 3-9 April 2021 1 4 2021 397 10281 12371237 © 2021 Elsevier Ltd. All rights reserved. 2021 Elsevier Ltd Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. ==== Body pmcRacist anti-Asian incidents and rhetoric in the USA have been on the rise during the COVID-19 pandemic, by some accounts increasing as much as 150%. The horrific mass shooting on March 15, 2021, in which six of eight people killed in three spas in Atlanta, Georgia, were Asian women, has prompted urgent conversations about prejudice against Asian Americans. Organisations including the American Medical Association were swift to underscore that racism, in addition to gun violence, is a public health crisis. The American Psychiatric Association warned that the shooting could compound the trauma and fear already experienced in Asian American communities. 2020 was a year of reckoning around race in the USA and national introspection about the maltreatment of people of colour; 2021 is a year to consider how racism and discrimination, alongside other social determinants, shape the broader context of health. April is National Minority Health Month and an opportunity to draw attention to Asian American health and wellbeing. The term “Asian American” carries a measure of controversy in trying to define an extraordinary mix of people, cultures, and languages. It is not a matter of semantics, but an issue of representation. Asian Americans include people with ancestry from east Asia, south Asia, southeast Asia, and in some instances Pacific Islanders or Native Hawaiians. Asian Americans constitute about 6·8% of the population of the USA and unlike any other ethnic or racial minority group, about two-thirds are foreign born and have entered the USA in the past 10 years. The largest Asian American subgroups are Chinese (4·2 million), followed by Filipino (3·6 million), Indian (3·3 million, and the fastest growing), Vietnamese (1·9 million), and Korean (1·8 million). As a predominantly immigrant minority group, Asian Americans can face specific barriers to accessing health care such as residency requirements for Medicaid eligibility (health coverage for low-income Americans) or being more likely to be employed in jobs that do not cover private insurance. Language proficiency can also limit an individual's ability to navigate a challenging health-care system. In relation to educational attainment and income level, immigration status can vastly bifurcate health outcomes. Providing a comprehensive picture of the health of Asian Americans is complex, challenging, and incomplete. Only in the past decade has research on health outcomes and disparities among Asian Americans gained momentum. Scarce data can routinely obscure or minimise health disparities for ethnic and racial minorities. In a Correspondence, Nancy Krieger and colleagues point to the stunning and continued paucity of ethnic and racial data being collected for COVID-19 vaccination in the USA. In general, Asian Americans have historically been seemingly healthier than other groups and compared with the US general population. For example, overall cancer incidence is lower in Asian Americans than in non-Hispanic White people. However, Asian Americans are at increased risk for liver and stomach cancers and are the only group for whom cancer remains the leading cause of mortality. Type 2 diabetes is more prevalent in Asian Americans as a group (9%) compared with non-Hispanic Whites (7·2%), but it is substantially higher in subgroups of Asian Americans such as Filipino men (15·8%). Understanding such differences could inform the use of prevention or earlier screening strategies. Tailored public health strategies to improve health equity will be an important means to counter the so-called model minority myth—ie, the expectation of excelling socially and academically. Asian Americans are often portrayed as self-sufficient and resilient, and are the least likely of all ethnic groups to seek mental health treatment. Cultural pressures and feelings of shame or stigma, especially around mental health disorders, treatment for cancer, and previous trauma can be deterrents to seeking help. However, culturally competent care could be improved through some of the solidarity and impact of Asian Americans as health-care providers. About 17% of all US physicians identify as Asian American. Although over-represented in number, increasing visibility and obtaining more leadership positions within health care is an important goal. Health equity in the US demands recognition of the contributions that immigrants make to society, understanding and provision of appropriate responses to the different needs of groups and individuals, and the dismantling of racism and discrimination against Asian Americans. For more on racism-related stress from the model minority myth see https://www.apa.org/pi/oema/resources/ethnicity-health/asian-american/stress-racism © 2021 RyanJLane/Getty Images 2021

33812475

PMC9753908

NO-CC CODE

2022-12-16 23:26:21

no

Lancet. 2021 Apr 1 3-9 April; 397(10281):1237

utf-8

Lancet

10.1016/S0140-6736(21)00769-8

oa_other

==== Front Organ Dyn Organ Dyn Organizational Dynamics 0090-2616 0090-2616 Elsevier Inc. S0090-2616(20)30054-1 10.1016/j.orgdyn.2020.100802 100802 Article Five Steps to Leading Your Team in the Virtual COVID-19 Workplace Newman Sean A. Ford Robert C. 31 10 2020 January-March 2021 31 10 2020 50 1 100802100802 © 2020 Elsevier Inc. All rights reserved. 2020 Elsevier Inc. Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. The emergence of COVID-19 has presented employees and employers new challenges as many employees and managers were forced to work in a remote environment for the first time. For many reasons, managing virtual teams is different than managing employees in a traditional face-to-face office environment. Although many managers have been learning how to lead their virtual teams over the last several months, we offer five steps for leaders to follow for how to maximize the effectiveness of a remote workplace. By taking specific actions and ensuring the organization has a culture to support their virtual workforce, leaders can improve the performance output and engagement of their teams. The five steps are: first establish and explain the new reality; second, establish and maintain a culture of trust; third, upgrade leadership communication tools and techniques to better inform virtual employees; fourth, encourage shared leadership among team members; and fifth, to create and periodically perform alignment audits to ensure virtual employees are aligned with the organization’s cultural values including its commitment to mission. All these steps start with the realization that managing a team is going to be different when the members are dispersed, and new leadership strategies, communication routines and tools are required. ==== Body pmcAfter Fred finished his fifth Zoom meeting of the day, he sat for a few minutes and reflected on this new world of managing. His team was working from home and so was he. The face-to-face contact that was a key part of his leadership style was now denied to him and it looked like it would continue to be unavailable for some time to come. The question he was facing was the same one many thousands of other managers like him were asking: “How can I be effective as a leader of a virtual team while keeping my team safe in this COVID–19 world?” The good news is that over the last 20 years, there has been a trend of employees moving to increasingly virtual work environments and much has been learned about how to lead virtual teams. This trend has grown exponentially with the emergence of the COVID-19 pandemic when many organizations went from having a modest percentage of team members working virtually, to the entire staff working from home. A 2020 survey of 2,865 employees by Global Workplace Analytics found that 67% of those surveyed in the U.S. were working from home for the first time due to the COVID-19 pandemic. As an indication of how well this was working, the same survey reported that only 19% of the respondents wanted to continue working from home full time in the future. Clearly, the transition has not been a smooth one for many. Previous studies of virtual teams have documented how challenges for both leaders and employees working remotely can be daunting. However, challenges can be more daunting for both leaders and employees who have had to suddenly shift their work patterns from an office to a home environment. The suddenness of this change made it difficult or impossible to adequately prepare leaders to lead in a virtual work environment. Many suddenly found themselves needing to ensure that their employees had access to not only obvious things like support systems, reliable internet, appropriate computer interfaces, remote access to firewall protected databases or even a quiet place to work at home but, also, to less obvious things like learning Zoom or other group collaboration technology in order to hold impromptu brainstorming sessions, conduct interactive meetings to problem solve, sustain the culture, and enable informal discussions. When the face-to-face interactions that helped bond a team all became impossible they became challenges for leaders to address. A survey taken since the emergence of COVID-19 revealed that only 46% of respondents had access to basic collaboration technology, 64% had no remote work policies, and only 41% had a clear understanding of their role and priorities. These data illustrate a new major strain on organizations and managers who may have had little experience or training on the unique challenges of managing remote employees. For many reasons managing virtual teams is different than managing employees in a traditional face-to-face office environment. While there are some benefits to working virtually such as better work-life balance by working from home, more efficient use of time gained by not commuting to an office, and increased access to the best talent that can be located anywhere, there are also unique challenges faced by virtual employees. Employees may feel lower levels of trust with and support from their manager and their organization as a result of working remotely. Moreover, since strong cultures are created by interactions with others in that culture and the visible reinforcements of cultural values found in an office’s signs, symbols, and artifacts, working from home inevitably diminishes the employees’ connections to the corporate culture’s values, beliefs and norms. Perhaps most importantly, the loss of frequent informal communication creates the need for leaders to employ new communication tools and techniques for their virtual employees. While few organizations were prepared to help their many managers cope with how to be effective leaders in this new reality, there is help available from research done over the last 20-years on leading virtual teams. Insights have been gained on how to lead, organize, motivate and build organizational support systems and strong cultures which enable virtual employees to be successful. The purpose of this paper is to organize and present strategies that have been found successful by organizations who have already met the challenges of leading virtual teams. We suggest that these strategies can be employed by the many organizations and managers that are now facing the challenges of leading productive employees in a COVID-19 world. The goal is to renew the work team’s culture in the new normal. Organizations and leaders know the critical importance of culture. It is the “software” that defines the values and beliefs of an organization as to how its members work with each other. Culture creates a work environment where leaders align their team members’ personal goals with those of the organization through organizational support policies and processes, formal and informal communications, and the leader’s behaviors. In the face of the major work environment changes that the COVID-19 pandemic has created, reestablishing and renewing the work culture with its shared values and beliefs has become an even more important leadership task. While the world of work has changed, the need for leaders to maintain their organization’s strong culture has not changed. However, the actions and activities by which they do these things has. In other words, they must take the steps necessary to renew their team’s cultural values, beliefs and norms by adapting to the new reality. The importance of this renewal to sustain a strong culture is especially important for virtual employees because a strong culture can help substitute for the lack of in-person communications and serve as a catalyst for strong communications across and within the team regardless of where the employees are located. If the leader’s pre COVID leadership style depended on face-to-face communications and personal interactions, the new reality of managing virtual employees requires leaders to learn and use new communication tools and managerial techniques to sustain the culture. We offer five steps for leaders to follow for sustaining and reinforcing a successful culture in a workforce that includes employees working virtually either full or part time. These steps address how leaders can help employees successfully confront the challenges they face when working remotely from home. There are things the organization can do and that leaders should do to ensure that the now virtual employee can continue to be productive as a part of team that must interact with others in new ways. These things range from establishing new organizational technical and employee support systems to leadership training on how to not only effectively communicate with virtual employees but, also, how to sustain their cultural values, beliefs, and norms in a trusting relationship with team members and the leader. To echo the consistent findings of Gallup surveys, employees need the tools to do their jobs and the feeling that someone cares about them and their progress at work. These are hard enough for leaders to achieve in a face-to-face work environment but even more challenging in a virtual one where tech support, leader feedback, and collaborating team members are far away. The five steps we offer are, first to establish and explain the new reality; second, sustain the corporate culture and reinforce the perception of leader trustworthiness; third, upgrade leadership communication tools and techniques to better inform virtual employees; fourth, encourage shared leadership among team members; and fifth, to create and periodically perform alignment audits to ensure virtual employees are aligned with the organization’s cultural values including its commitment to mission. All these steps start with the realization that managing a team is going to be different when the members are dispersed, and new leadership strategies, communication routines and tools are required. Step 1: Establish and explain the new reality The first step is the get the team to acknowledge that this new reality represents a change and all the fears and anxieties that people feel in the face of change will be addressed by the leader. Leaders should be transparent about any changes the organization may be taking regarding business strategy and product or services be offered which may impact work activities. Change means uncertainty and the leader should access all the tools available for managing change that have been used successfully in the past. Gaining employee acceptance that they must learn how to work at home is the first step a leader should take to be successful in this new reality. Understanding that this may not be a short-term phenomenon is also important to encourage employee acceptance of change. This new reality can’t be ignored and requires everyone to learn new communication tools, time management skills, and interpersonal interaction skills. People working at home are encountering multiple personal and professional issues that a leader needs to address to sustain their productivity. The new normal for the team requires the leader to display a greater sense of empathy and sensitivity to the challenges of working out of the office. The first question asked of a team member should always be, “how is the family or how are you doing?” Being quarantined can be difficult for families and a very lonely vigil for singles and leaders need to recognize this as much as they recognize the distractions of having family interruptions. Simply put, it is important for leaders to start conversations by communicating a genuine concern for the safety and well-being of the team members and their families. It is also important to acknowledge the uncertainties everyone shares over what the new normal will be as well as how it will affect each of them and the team. The leader must be able to tell everyone what is known about the organization’s policies and procedures for working in this new environment, the resources available to help, and any changes in the work flow and mission. This means, of course, that the organization should have policies, procedures and systems support for remote employees and, if it doesn’t, the leader should create operational procedures and policies that define expectations for those working from home for the first time due to the pandemic while actively advocating for the organization to act. Eliminating uncertainty is a key task for the leader and the team members need to know that their leader as the primary link to the organization is actively working on their behalf. Ramping up communications should be done in two ways. First, establish regularly scheduled one on one calls or video enabled meetings to establish regular contact with each direct report. This allows for open communication, expressions of personal concern for their welfare, active listening to any concerns or frustrations, and reinforcement of the culture, values, and mission of the organization. Leaders in these calls can acknowledge their awareness of the unprecedented pressure their employees are experiencing while staying at home, teaching their kids, missing social interactions with friends and coworkers, wondering where they can find a peaceful place in their homes to work, and feeling nervous about themselves and their organization’s future. Even if the leaders cannot tell their team members all they want to know, leaders should tell them all they do know and demonstrate that they care about them, their well-being, their future, and their families. Empathy is important in uncertain times like these. Second, after establishing regularly scheduled one on one meetings to check in on their team members individually, the leader needs to schedule weekly team meetings. The team needs to be reminded of the importance of their continuing operations as a team. These meetings generally use video conference technology like Zoom or WebEx technology, but other options are available and may be more desirable depending on needed technology and security protocols. Regardless of the technology, such meetings should accomplish four important goals. First, to communicate and reinforce the organization’s goals and how the team’s goals connect to those goals. Second, that the organization has, is getting, or will sustain the technological resources needed to support the team members at home. Third, that the leader is committed to sustaining the team and its members, its culture, and its ability to achieve its goals. Fourth, how the team is doing in reaching its goals (the performance metrics) and what problems the leader must resolve. Accomplishing these meeting goals has several important elements. The first element is to establish ground rules for communication with the team members, across the team, and with the team’s internal and external customers. A communication etiquette should also be established; time limits on responding to team member questions; setting times for interaction that respect family obligations and time zone differences, and participation expectations for discussions that require personal interactions instead of relying only on email and texts. Responsiveness is very important when employees are working virtually. Besides providing necessary information to a team member, a leader’s timely response also sends a strong message to a distant employee of respect - that the leader is paying attention and doing whatever is possible to support that member and the team. The weekly team meetings should also have a rhythm of a set time and length as if the team was back in the office such as a one hour meeting every Monday morning at 9. The time set should show recognition of family obligations or other commitments associated with working from home. A second weekly meeting should also be established for the team members to collaborate and engage with each other to discuss work or any other topic they wish to discuss. In a traditional office environment many issues and obstacles to task completion are resolved by face-to-face encounters among and between team members. When these interaction opportunities are eliminated, an alternative process for collaboration should be created such as online chats and screen share applications like Skype or Lync. Leaders should encourage online collaborations, so they can happen organically or in an “on-line break room”. Leaders should also periodically participate in these to keep track of issues and concerns needing intervention, further discussion, or involvement of other organizational units. To replicate some of the camaraderie and informal interactions of face to face meetings, the leader might even introduce some fun activities like warm up “do you know” type quizzes, virtual pizza/ice cream celebrations, or other techniques to build sense of community and team collaboration. When teams must collaborate, facilitating it becomes a critical concern for leaders when employees are working from home. Both task and informal collaborations allow everyone to stay connected to a mission driven culture and leaders should reinforce the norm that team members are expected to interact with each other to help solve problems, share knowledge, and resolve issues even when they are not able to connect face-to-face. Leaders should reinforce how important this continuing contact is to team success and publicly recognize and reward positive team collaborations to reinforce the behavior. The more interdependent the team’s tasks are the more important this collaboration is, but it is still an important leadership responsibility even for largely independent workers. Meeting often with their colleagues virtually reminds everyone that they are part of a team and a shared culture. Communicating the new reality through establishing new routines to enhance communications with now virtual employees is important to establish the new reality. In a typical office work setting there are routines with established patterns of communication and interaction. Without the office’s structure, the leader must create routines that can substitute for that office structure. Moreover, leaders can add new options to enable employees to use any extra time for personal and professional growth or just to promote mental health in stressful times. Hewlett Packard, for example, started offering meditation and mindfulness services from Headspace to help employees unplug from work while at home. Other companies offered their newly remote workers on line access to yoga and Pilates classes or professional development, educational, and skills-based programs. Such routines ensure that the team members stay focused on accomplishing team goals while acknowledging the additional challenges team members face when working at home. Managers successful in this step place extra emphasis on their goal setting, and performance management skills. These become more critical for virtual teams as individual and tasks need greater task structure, defined project time lines, and measurement of specific goals. Setting defined goals instead of defined work hours allows the team members the flexibility they need to accommodate their new realities. When task goals are specific, measurable, actionable, realistic and time limited (SMART), employees can use nontraditional work times to complete work tasks and better accommodate any family time demands that occur while working at home. While goal setting is an effective managerial tool in traditional office settings, it is critical in virtual settings where goals should be set for performance, behaviors and learning. With specific goals the leader can focus on the metrics that matter for each team member and the overall team. We offer the following leader actions to implement this step:• Hold weekly meetings with entire team early in week at a set time that accommodates team members’ at-home responsibilities and obligations to check on goals, progress, and problems. • Hold weekly meetings with each team member at convenient times for them to review progress towards goals and to identify any personal, professional or team problems that need leader resolution. • Schedule weekly meetings for team to collaborate and build relationships • Start meetings asking two questions; “How’s everyone doing?” and “Are you having any problems I can help you with?” • Remind team of its importance to overall organization and other stakeholders in their work output. • Define and enforce communication norms and etiquette Step 2: Sustain the corporate culture and reinforce the perception of leader trustworthiness After communication norms and work routines have been established, the second step is to establish an environment that sustains and reinforces the team’s commitment to the organization’s culture. As virtual team members are extra dependent on the leader to define and sustain the organization and team culture, the leader needs to spend extra time and effort building trustworthiness. Employees need to trust in the leader as truthteller, source of organizational and team knowledge, and active advocate to the organizational leadership and other units that control resources needed by the team. Building and sustaining trust is especially important for managers of remote employees where there is a high risk of team members feeling isolated. Having lower levels of trust can negatively impact their engagement in the culture, the organization’s mission and team productivity. There are many benefits when there is a high level of trust in the workplace and the leader’s effective management of the culture is critical. The research shows that trusting team members are more proactive, have a higher level of focus on task output, display more optimism, communicate more often, and are more open about providing feedback. To sustain the culture and build leader trustworthiness, organizational leaders should make sure they have learned how to employ the tools of establishing culture. Leaders teach culture by what they consistently say, do and write. Since culture is shared the routines established in Step 1 help ensure that the team members have a time to share it. The leader should add to the words defining the culture by using visual reminders such as superimposing a values statement on the web screen, using it on a background for virtual meetings, or even sending corporate value statements and symbols to employees for display in their homework locations. Celebrations of important personal and professional milestones remind team members of what is important to the culture and each other even if they can’t do it in person. Asking team members to comment on how the team values came into play in each job or reading correspondence from stakeholders can remind team members that even though they are at working at home they are still part of something important and worthwhile. In other words, the same effort to establish team culture and how it fits into the overall organization’s that was used in the office should be the minimum effort for leading a virtual team. It is harder to do so it will take more effort and creativity by the leader. Other things the leader can do to established and sustain a trustworthy culture can be divided into four categories: advocating for organization infrastructure and policies that set team leaders up to be successful, leadership actions that ensure they consistently follow through on commitments, and decisions that treat employees fairly based on transparent, objective policies, and, finally, team trust building. Organizational Strategies. The strategies an organization adopts to support virtual employees establishes the infrastructure and culture for the organization to earn its employees’ trust. Organizational trust is earned through thoughtful policies and procedures which take into consideration the unique needs of their virtual staff. This can be done through ensuring the adequacy of the technology infrastructure and tech support, training leaders to succeed in a virtual work setting, ensuring team tasks, roles, and expectations are clearly defined, and having human resource policies aligned with the needs of virtual employees. Through careful attention to these organizational responsibilities, employees feel they can trust their organization to support their virtual communication needs as well as feel respected, recognized and supported by organizational policies and processes. The leader should actively advocate for these and make it known to team members that the leader is trying to ensure that the organization is acting in trustworthy ways. The first issue for organizational attention is the critical one of technological support. Newly distant employees need quick resolution to the technological issues that working from home creates. Timeliness signals organizational concern and builds trust of distant employees that the organization cares and will do what it can to solve these frustrating issues. Technology hardware, software, applications, and tech support need to work properly if the organization expects employees to be efficient and effective. What is generally available in an office is more critical for the distant worker sensitive to any signal that the organization is unconcerned with those working at home. The virtual leader should do everything possible to make sure the organization is responsive to the technology needs of each employee. The new work at home workplace can have data security and privacy issues not faced in an office setting. The organization will need to provide training and instructions on how to access the company’s technology and data base from their home devices. In addition to the learning how to access the organization’s data and systems, employees will need to be taught the enhanced collaboration tools being used in order to enable employees working remotely to collaborate effectively, share their desktops, and have impromptu meetings. Tools like Zoom, WebEx, Microsoft Teams, and Slack allow employees to voice and video conference or collaborate through instant messaging of questions or notices. This can be done between two employees or within a group to create a virtual chatroom. These tools allow employees to share their screens to collaborate on a shared document or troubleshoot a task problem. Software applications like Microsoft OneDrive, SharePoint, Drop Box, or Google Docs allow employees to concurrently work on data sets or projects without access or replication issues. By providing a thoughtful, comprehensive technology infrastructure and training, employees working remotely will feel the organization cares about their ability to be successful in their new work from home arrangement. Much of the technology that supports employees working virtually sits on cloud services and only requires internet access to utilize these applications. However, organizations should also prepare for and communicate contingencies for when employees lose access to the internet, or when support applications go down. Companies should consider having the ability to rapidly deploy air cards overnight to employees who may be without internet service for more than a short time. Organizations should also ensure that processes are documented and trained back-ups available, so others are able to step in for those encountering technological challenges to complete time sensitive tasks. In addition to technology cues which help create a supportive, trusting culture for virtual employees, the organization should ensure the leaders are trained and responsible for spending the extra time and effort in defining task and role expectations for both team and individual team members. This means all employees should understand that each has a role to play in the team and exactly what that role is. This can be communicated in several ways. First, every job should have a job description that clearly defines key responsibilities for each employee. The job descriptions would then be supported by documented operating procedures (desk procedures) for common tasks which establish the responsibility of each person on a team and the timing that each task should be completed. Next, work plans of who is assigned to specific tasks, with the task priority, and due dates should be documented in a place where everyone on the team can access and review the project plans. Last, regularly scheduled team meetings should occur so that virtual team members can share their status, risks, and collaborate on solutions. All these activities create a role clarity and consistency which will create a trusting environment. Human resources also play an important role in building and maintaining a trusting culture for virtual employees through the programs it creates for training, rewarding, and recognizing employees as well as the policies it creates for the unique issues associated with employees working at home or at distant locations. These policies and procedures should account for what employees need now as well as plans for their future. If this pandemic exists long enough, then HR will need to create new policies and procedures for handling recruitment, selection, and on boarding of new employees in the virtual workplace as turnover will inevitably occur. HR will also need to review all its plans and procedures for handling employee issues such as promotions, career counseling, discipline, training, drug testing, and leadership development as they will all require different procedures in the new reality. Moreover, the unique issues of work-family conflicts, mental health, and work stress will necessitate new ways of recognizing and resolving these employee issues when they are not in an office. New training for leaders to check on how employees are doing will be required to ensure that mental health issues are caught early, and resolutions offered in a timely way. Distant workers will need new ways to access employee assistance programs, career development counseling, and personal disputes that many organizations will have to invent as they never had a virtual workforce before. The way in which HR deals with the newly virtual employee will send a strong signal to all employees as to how trustworthy the organization is. Indifference to the unique challenges many face when working from home is a path to destroying trust in organization. Additionally, ensuring appropriate training is available for virtual employees is an important role for human resources. Training should be made available on demand as modules, so employees can access it as needed. This training can range from job skills related to professional and career development (e.g., executive education and academic programs) to personal health (e.g., yoga and Pilates) but, the organization’s provision of this training signals its concern for the development of its employees regardless of location. Beyond training, human resources should make sure policies related to pay and rewards are adjusted to account for a virtual workforce. When employees are no longer in the office, greater transparency is needed to communicate to all how the team and its members are tracking towards performance goals. A trustworthy leader is one that reminds HR that high performing team members should be in consideration for promotions or as candidates for further training. The routine meetings managers have with the team need to spend time addressing progress to goals and also recognizing success. These meetings should also enable leaders to manage performance issues. While in the office, this may have happened in quick, impromptu coaching conversations or feedback, managers now will need a forum and scheduled frequency to provide feedback. Human resources can play an important role in reinforcing this element of the culture by institutionalizing what metrics will be used to measure performance, where they can be reviewed, and guidelines on performance conversations that managers can leverage. They can also encourage managers and leaders to celebrate successes. Lastly, human resources should establish consistent policies related to employees working from home. These policies should include what expenses (internet, additional phone lines, office supplies, etc.) can be reimbursed. In addition, whether or not there will be any changes to core working hours and what to do if an employee goes missing for longer periods of time during the normal business day. By establishing working at home rules, human resources can establish consistent standards which when equitably applied to all employees, increases trust. Leader strategies. Leaders play the most important role in building trust. Effective virtual team leaders have learned to adapt their communication and management approach to better respond to the needs of a virtual employee. Virtual employees may live in different time zones or have different work environments that a leader must accommodate in how and when to communicate with those employees. Distant employees also no longer have easy access to informal hallway conversations where information and concerns are readily shared. Leaders who understand and address these needs for information and actively work to fill the communications gaps in communications build higher levels of trust. Employees will see them as important and honest sources of the information that used to be available in the hallways and that provides important contextual supplements to the traditional work-related communications that all leaders provide. In addition, leaders need to recognize that there may be less communication regarding the organization's culture, strategy, and goals when employees are working from home. Visual cues in the office are no longer available, so leaders must be thoughtful for how they provide organizational updates. The importance of these should not be neglected just because they cannot happen in person as they had in the past. Organizations’ business updates or cross organizational information sharing are critical vehicles to keeping employees aware of and informed about company’s values and progress towards goals. Employees not routinely provided with this information will struggle with maintaining the organizational commitment and cohesion that sustains a culture and a commitment to an organizational mission. Effective virtual leaders listen well to hear the issues their team members raise, to catch the nuances in the conversations, and observe how comments and texts are phrased. Without the informal and nonverbal communication that leaders rely on in the office setting to complement the verbal, virtual leaders must listen extra carefully as the nonverbal cues are no longer available. These are how leaders sustain a trusting culture. Team Strategies. The last important area of focus for a leader to address in order to sustain a trusting culture with virtual employees is ensuring that team members can see how they collectively and individually connect to the organizational strategy and goals. Similar to other organizational and leadership strategies to accommodate virtual employees, managers should over communicate (e.g. frequent status meetings, sending out updates to the team, having a publicly accessible project plan on a tool like SharePoint). Within these meetings, having leaders ask for feedback and ideas further engages team members into accomplishing common goals. Leaders must show by their words that they want team members to feel psychologically safe. Feeling safe so employees will continue to take appropriate risks, speak up with their concerns, and continue to take initiatives on new projects and opportunities is harder to do in a virtual world but no less important. Leaders should enhance their willingness to listen with empathy and speak with sincerity as the best available tools they have to communicate that they retain their faith in their employees’ judgment and capabilities even when they are unable to physically pat them on the back or do the other things that are done in office settings. Leaders know the importance of ensuring team members of their psychological safety. Psychological safety is often defined as a condition in which human beings feel (1) included, (2) safe to learn, (3) safe to contribute, and (4) safe to challenge the status quo – all without fear of being embarrassed, marginalized, or punished in some way. The research shows that leaders who promotes an environment that is psychologically safe have teams with higher levels of success and reinforce trust in the leader. In the new reality of virtual organizations, the leader may become the organization to team members and the role of leader in establishing and sustaining a psychologically safe work environment is critical. We offer the following leader actions to implement this step:• Find ways to remind team of cultural values, beliefs, and norms in what is said, done and written in both team and individual communications • Use rituals, virtual celebrations of professional and personal milestones, and send physical symbols of the culture to employee homes to keep culture top of mind for team members • Remind team members of communication and data sharing protocols that build trusting relationships • Don’t forget to include those who remain at home during partial reopening of offices • Engage with tech unit to ensure team members have ready access to tech support and best available collaboration equipment and platforms • Engage with HR to ensure team members have ready access to professional development training, educational programs, career counseling, pay and benefits, mental health support (e.g., EAP, Yoga, etc.), and corporate recognitions and communications. • Engage with other organizational units and customers of work product to ensure team have ready access to needed resources, information, and cooperative collaborations • Remind team members that the new work environment is still psychologically safe as they (1) are all included, (2) safe to learn, (3) safe to contribute, and (4) safe to challenge the status quo – all without fear of being embarrassed, marginalized, or punished in some way. Step 3: Upgrade leadership communication practices and techniques to better inform virtual employees Although communication is implicit in much of the earlier discussions of steps, it is so important it merits consideration as its own step. Leaders should be trained to understand that communication techniques which may have worked well in face-to-face settings need to be modified or enhanced to meet the communication needs of virtual employees. Christopher Reynolds, Chief Administrative Officer of Toyota North America stated in a recent Wall Street Journal interview on managing virtual employees, “There’s no such thing as too much communication.” Obviously one of the key factors in using virtual teams has been the development of advanced communication technology to ensure adequate communication with distant workers. However, just buying and distributing communications technology to virtual employees will not automatically make them effective. Leaders must also find new ways to compensate for the lack of nonverbal communication cues that are generally available in an office environment. As well, virtual employees do not benefit from the informal, water cooler talk where employee bonding and information sharing often occurs. People working away from the office have often felt left out of the communication flow anyway and, if the leader not actively fill in those gaps, there is likely to be misunderstandings and confusion. Communication gaps can lead to lower employee engagement, lower levels of team and cultural cohesion, and reduced trust between and among employees. The reality for newly virtual employees s makes these gaps more apparent and leaders must expend greater effort to align employees with the organization’s culture as well as its goals and objectives. To eliminate the gaps and reduce their adverse outcomes, leaders must adapt their communication practices and techniques for their virtual employees. Researchers have found that specifically focusing on communication frequency, predictability, responsiveness, clarity, and mode can help overcome the challenges of working with virtual teams. Paying attention to these communication tools and techniques helps ensure employees working from home feel connected to the organization's mission and culture. First, regarding frequency, frequent leader communication creates a stronger relationship with virtual employees and more communication between and among employees. Indeed, a recent Gallup survey found that employee engagement is higher when managers communicate with their employees on a daily basis (e.g. email, text, instant message, etc.). Frequent communications, formal and informal, increase team effectiveness and performance. For example, a leader may have scheduled weekly team status meetings and monthly one-on-one meetings with people on their team with group chats or texts in between to ask team members about their weekend or make sure they saw an important email. Increasing frequency could also be accomplished with a weekly blog, newsletter, or business social media application like Yammer or Slack. Leaders could also use instant messages to regularly check in. A second communication technique that is important for leaders is the degree to which their communications are predictable. Predictable responses to inquiries or task requests positively impacts a member’s organizational commitment and team performance. Leaders who provide detailed, thoughtful responses build trust and improve a member’s sense of engagement and feeling of being valued and respected. Thus, a leader should avoid being unpredictable as it can cause a team member to read something unintended into the variance in communication (e.g. you are upset with their question, you do not find their communication very important). Such unpredictability in communication breaks down trust and can change an employee’s perspective on the culture and leader’s trustworthiness. A third related communication technique that needs the leader’s attention is timeliness of response. Research by Gallup finds that employees expect calls or messages to managers and leaders returned within 24 hours. The expectations on how quickly a manager should respond to texts is even quicker. The more timely the communication back to virtual employees is, the more engaged and committed employees are to the organization and attaining specific goals and objectives. The expectation of timely responses does not also mean that a manager must provide a comprehensive or complete response to a complex question as the sender knows that research or follow-up with others may be required. A good best practice for timely communication is immediately or as soon as possible acknowledge receipt of the email and include in the response a time line for providing a complete response. This response builds trust as it signals the recipient that the inquiry and the sender are important and deserve the leader’s quick attention. The fourth communication technique that leaders should attend to is clarity. Leaders are expected to send communications that are clear. Whether the communication is giving direction or feedback, the communication needs to be understood by the receiver. Indeed, a classic definition of authority is a leader’s communication that is accepted by the recipient. Implicit in this definition is that the communication must be understood. Employees cannot do what they don’t understand so communications must set clear direction and expectations. Clearly documented task assignments, measurable goals, team member roles, and due dates positively affect team performance. In a traditional face-to-face office setting, much performance feedback or coaching on tasks happens organically and informally at employees’ desks, in break rooms, or quick, impromptu huddles, or sometimes in just a leader’s simple facial expression or gesture. In a virtual workplace, the informal and nonverbal communications are largely gone so communication must become clearer by being formal and documented. Lacking the informal and nonverbal options, managers and employees talk on the phone, instant message, or meet online. Because of the limits of those communication techniques, managers of virtual employees must follow-up conversations with clearly documented summaries of expectations. Likewise, results from online huddles must be clearly documented so that everyone understands each other’s task responsibilities, priorities, and due dates. Providing clear roles, goals, and task expectations also helps align virtual team members with the organization's culture, mission, and vision. and with each other. The last communication technique that needs special attention by virtual leaders is thoughtful selection of the mode of communication to ensure a best fit of the message with the recipient. Selecting when to use an email instead of a phone call, or instant message, depends on how the recipient expects to receive that particular message. In effect, the recipient is the leader’s customer for a message and the leader should consider the way in which each recipient prefers to receive different types of information when selecting the best mode. Many leaders have used video calls with Skype, WebEx, or Zoom but they are not always the best fit for every communication between a leader and the team. If documentation is needed as a reference for specific work tasks, tracking decisions, or technical analysis, then the mode that allows that is the best fit. For urgent notifications or requests, instant message or text is generally most effective. If something is very important leaders should use multiple modes to ensure the redundancy that increases the odds that communications are received. Multiple mode usage also signals urgency. Zoom or WebEx may work best for l team meetings where people are familiar with each other or small group meetings where initial introductions can benefit from the intimacy. However, for other meetings a standard conference call with screen share (to show or collaborate on documents) may be the better choice. The choice of communication mode is very important because it impacts how virtual team members interpret the message. Overall, these communication practices and techniques should be considered together as each contributes to the effectiveness of the communication in reaching the virtual team members. Managers who are able to adapt their communication frequency, predictability, responsiveness, clarity, and mode to be effective in the messages they send to their virtual employees will benefit from improved performance and increased trustworthiness while increasing their ability to effectively communicate goals, culture, and values. In addition, these communication practices and techniques provide the structural support for leaders to effectively develop their employees and manage through performance issues. The bottom line is that when an organization transitions to a virtual workplace, it puts an extra burden on the leader’s ability to communicate effectively when the physical reminders of the company’s culture and mission and the informal communications that are part of the traditional office environment are no longer available. We offer the following leader actions to implement this step:• Communicate in a timely way to show respect for team members and responsiveness to their needs (responding quickly sends a message so check all modes frequently for messages) • Communicate frequently with team members to show they aren’t being forgotten • Communicate in predictable patterns to avoid feeling of uncertainty • Communicate clearly by checking to ensure that what was communicated was understood • Carefully select the communication mode that best fits the message and the receiver • Monitor closely the technology to ensure communications are smooth Step 4: Encourage shared leadership among team members Following the steps to earn trust and communicate effectively with virtual employees’ managers can now focus on the method of leadership that is most effective for virtual employees. Shared leadership is the process where team members each play a role in the collective leadership of team tasks. Research has found that shared leadership can improve performance results from virtual teams when compared to more traditional leadership techniques. For virtual teams, shared leadership can take on greater value because some of the benefits of shared leadership are directly related to challenges in virtual teams. Specially, shared leadership requires greater team member engagement and group interaction, which leads to team members building more cohesive interpersonal relationships and working together more closely. If team collaboration is a critical part of virtual team success, shared leadership is an important way to enhance that collaboration. To successfully implement shared leadership within a group of virtual employees it is recommended to start with a training strategy to help members understand how shared leadership works and what successful team outcomes look like. Shuffler and his colleagues have discussed the culture and infrastructure that should be established to enable successful shared leadership on virtual teams. The infrastructure should provide a process for how team members will coordinate with each other to provide risks, project needs, and status updates. They suggest establishing a group mission, expectations and goals, and then create a defined process for team members to collaborate, provide updates, and share feedback. Leaders would then shift to a role of supporting and advising the team members on leadership roles, and monitoring and coaching performance. Since one of the challenges with virtual teams is the diminished ability of management to monitor employee activity and team dynamics, effective shared leadership can help meet this challenge. When implemented well, shared leadership changes the leadership paradigm. Instead of relying on the formal leader to provide all guidance, supervision, and performance feedback, shared leadership places virtual colleagues in different leadership roles with defined responsibilities for team outcomes. These roles increase member vesting in team outcomes. Moreover, shared leadership quickly exposes team members who are highly engaged by this level of ownership and those who are unable or unwilling to take on any leadership roles. Shared leadership can be especially important for virtual employees because it shifts the responsibility for maintaining teamwork and communication to the team members acting as a team instead of relying entirely on management to provide this collaboration. For example, a manager at a large outsourcing organization distributed the management of processing responsibilities to each member of the team. The entire team is collectively responsible for processing human resource questions and transactions which come in from other employees. These are grouped into different categories, such as, payroll, health insurance, retirement, disability and leaves. The manager has appointed one member of the processing team to be accountable for each category of processing to ensure that requests are completed on average in less than two business days, that customer satisfaction with the transactions is high, and that the accuracy of the processing is above 98% correct. Each week at a team meeting, the person accountable for each category of processing provides a status update and a report on any unexpected challenges, such as higher than anticipated volume. In addition, at the meeting, they share ideas for continuous improvement for their process from the review of the prior week’s results. Even though the team is working all requests collectively, this shared leadership allows for each team member to have direct ownership of a process and engagement is at a level deeper than if the manager was managing each transaction at the manager’s level. Executing a strategy of shared leadership has a great number of advantages for virtual teams and managers can benefit from thinking creatively as to how they might distribute ownership to engage more of the team in leadership roles. We offer the following leader actions to implement this step:• Train team members on how to take on leadership roles • Identify the leadership skills of each member to ensure proper assignment of roles that each is good at and which has development potential • Encourage team collaborations and information sharing to allow members to see each other’s contributions to the work product • Celebrate accomplishments of team and spotlight contributors • Ensure the infrastructure for work collaboration, providing updates, and sharing feedback is accessible, understood, and working • Monitor the team collaboration in routine meetings (Step 1) to ensure conflicts are identified and resolved • Coach those who need help in performing their leadership role Step 5: Create and periodically perform alignment audits to ensure virtual employees are aligned with the organization’s cultural values including its commitment to mission The last step is to create and periodically perform an audit to ensure that the virtual employees are aligned with the organization’s cultural values and its mission. It’s not enough to tell everyone that the organization has a culture and a mission. It is necessary to check. The purpose of an audit is to check on whether or not something is happening the way it is supposed to. This step requires a leader to do exactly that. Thus, we audit to find out if the virtual team members are aligned with the culture, the organization’s mission, the team’s tasks, and the leaders’ ability to lead. One critical key to building and sustaining trust is that everyone agrees on the team tasks, the shared culture, and the organization’s missions. If everyone agrees on what they are supposed to do, why they do it, and how they work as a team, the odds are very good that the leader has practiced the leadership skills that work best in a virtual team setting. By practicing the communication practices that best fit a virtual team, that leader has learned how to adapt the leadership skills to a virtual setting. A common belief is that what gets measured gets managed and the audit can be the metric on how well the virtual team is adapting to the new reality. We offer the following steps to perform an audit. It is short, focused on communication and caring. It balances the tasks of setting goals and assessing performance to keep the team and its members focused on the task with the leader’s practices of caring. Not only does it check to ensure that the leader has taken the time to present clear, operational goals in ways that the team members understand but, also, makes the effort to provide feedback on task performance for both individuals (privately) and team (publicly). The audit should capture the human side of the leadership as well as task performance and remind the leader to pay attention to the mental health aspects of those on a virtual team. There are issues for a leader to address for those employees that are out of sight and heard only via voice connections, emails or texts. The virtues of management by walking around are largely lost to the virtual team leader so the opportunities to catch the frown of dissatisfaction, to hear the grumbling at the water cooler, or to offer some coaching in an informal meeting in the hall are unavailable. Nonetheless, the dissatisfactions still occur, the reasons for the grumblings still exist, and the need of an employee to have a quick informal chat with the boss doesn’t disappear. The items we include on Table 1 can be added to or deleted as desired to fit the leader’s circumstances. It is a start to develop a tool that will force the leader to periodically review the key issues in building and sustaining the relationships with the virtual team members that yield a trusting relationship. It is the leader’s task to ensure that those who are unable or unwilling to be in an office environment still gain all the benefits of being part of a strong mission focused organization in which everyone is enabled and encouraged to perform their best work.Table 1 Team Alignment Audit: activities to ensure the leader has sustained the culture and aligned the virtual work environment to the organization. Leaders may decide to add or delete items based on their organizations goals and objectives Table 1Communicate New Reality Hold weekly team meeting early in week at a set time that accommodates team Hold individual weekly meetings with each team member Schedule weekly team meetings with the purpose of the team talking and engaging Start each meeting with two questions; “How are you and your family doing?” and “Are you having any problems I can help you with?” Remind team members of their importance to overall organization and other stakeholders in their work output. Define and enforce communication norms and etiquette Trust Find ways to remind team of cultural values, beliefs, and norms in what is said, done and written in both team and individual communications Use rituals, celebrations of professional and personal milestones, and symbols to keep culture top of mind for team members Remind team members of communication and data sharing protocols that build trusting relationships Include those who remain at home during partial reopening of offices in team events Advocate for those unable to return to offices because of family responsibilities, preexisting health conditions, or fear when and if offices reopen Engage with technology unit to ensure team members ready access to tech support and best available collaboration equipment and platforms Ensure team members have access to professional development training, educational programs, career counseling, pay and benefits, mental health support, and corporate recognitions and communications. Engage with other organizational units and customers to ensure team ready access to needed resources, information, and cooperative collaborations Ensure team members are encouraged to share ideas, take risks, and provide feedback Communication Communicate in a timely way to show respect for team members and responsiveness to their needs (responding quickly sends a message so check all modes frequently for messages) Communicate frequently with team members to show they aren’t being forgotten Communicate in predictable patterns to avoid feeling of uncertainty Communicate clearly by checking to ensure that what was communicated was understood Carefully select the communication mode that best fits the message and the receiver Double-check the technology to ensure communications are smooth Shared Leadership Prepare team members to take on leadership roles Identify the leadership skills of each member to ensure proper assignment of roles that each is good at and which has development potential Encourage team collaborations and information sharing to allow members to see each other’s contributions to the work product Celebrate accomplishments of team and spotlight contributors Ensure the infrastructure for work collaboration, providing updates, and sharing feedback is accessible, understood, and working Monitor the team collaboration in routine meetings (Step 1) to ensure conflicts are identified and resolved When completing the audit, we offer the following strategies to implement this step:• Schedule a periodic audit that assesses the status of key actions identified from the prior steps • Make sure audit has included important actions to periodically assess and doesn’t include actions that can be assessed less often • Add items that are unique to the leader’s leadership style, team goals, member personalities, and relationships with members • Ensure that employees know the organizations mission and goals and how their job impacts the pursuit of the mission and goals Conclusion The buzz phrase ‘the new normal’ has been commonly used to describe the changes in the world since the emergence of the COVID-19 pandemic. One of the most significant changes has been the many employees who have suddenly found themselves working from home, who had never worked remotely in the past, and had not previously had training in or experience with working from home. The disruption caused by this dramatic change has created stress for both leaders and their teams. While many organizations and their leaders have now worked through the initial challenge of how to make working remotely a reality, many are now reflecting on whether or not their communication and management routines are helping their leadership get the best results possible from their newly virtual employees. Even if employees are able to slowly return to their offices, there is a high likelihood that many employees and organizations will continue to have significant numbers of employees working from home in the future. Leaders must understand how virtual teams are different and take the action steps like those offered here to be successful. The good news is we have already learned much about how to do this from leading employees who have successfully worked remotely for over 20 years. There are some proven leader actions and techniques which can make virtual work arrangements effective. We have offered the 5-steps derived from the 20 years of experience to help managers of newly virtual teams to be more effective. Not only do we define each step but offer specific recommendations for using them. Perhaps the most valuable is the last step, the audit, as it forces leaders to regularly monitor the things that can impact their team’s effectiveness. This audit, a sample which is provided above, can help a leader check that he or she has taken the time to communicate effectively, build trust, share leadership responsibilities, and present clear, operational goals and tasks in ways that the team members understand. If everyone agrees on what they are trying to accomplish, why that is important, and how each team member has a contribution to make, the odds are very good that the manager has mastered the leadership skills that work best in a virtual team setting. An audit also helps ensure leaders are capturing the human side of the leadership as the items we include reminding leaders to pay attention to the mental health aspects of being on a virtual team. There is an old saying that no one cares what you know until they know you care and this is both harder to do when team members are not operating in a face-to-face environment and more important to do using all the communication tools and techniques available. There are issues that need a leader’s attention for those employees that are out of sight and heard only via voice connections or tweets. The many benefits of management by walking around are largely lost to the virtual team leader so the opportunity to catch the frown of dissatisfaction, the grumbling at the water cooler, or the informal meeting in the hallway is unavailable. Nonetheless, the dissatisfactions still occur, the reasons for the grumblings still exist, and the need of an employee to have a quick informal chat with the boss doesn’t disappear. By dedicating time to understand the new challenges caused by working outside the traditional office, leaders can recreate their culture to fit the unique needs of a dispersed workforce for now and in the future. Credit statement This manuscript was jointly conceptualized and written by Dr. Sean A. Newman and Dr. Robert C. Ford. There was no funding received for this article. Further Readings For additional information on successfully managing and communicating with virtual employees, see Newman, S. A., Ford, R. C., & Marshall, G. W. (2020). Virtual team leader communication: employee perception and organizational reality. International Journal of Business Communication, 57(4), 452-473.Also see, R. C., Piccolo, R.F, and, Ford, L. R. (2016). Strategies for building effective virtual teams: Trust is key. Business Horizons, 60, 25-34. For more on shared leadership with virtual teams, please see Hoch, J. E., & Dulebohn, J. H. (2017). Team personality composition, emergent leadership and shared leadership in virtual teams: A theoretical framework.  Human Resource Management Review,  27(4), 678-693, as well as, Shuffler, M. L., Wiese, C. W., Salas, E., & Burke, C. S. (2010). Leading one another across time and space: Exploring shared leadership functions in virtual teams.  Revista de Psicología del Trabajo y de las Organizaciones,  26(1), 3-17. For additional insight on organizational alignment see Alagaraja, M, & Shuck, B. (2015). Exploring Organizational Alignment-Employee Engagement Linkages and Impact on Individual Performance. Human Resource Development Review, 14(1) 17–37. Senior Vice President, Aon, MBA Instructor University of Central Florida, Rollins College, 337 Swansea Court, Oviedo, FL 32765, 407-902-5526t. E-mail: snewman@rollins.edu

0

PMC9753911

NO-CC CODE

2022-12-16 23:26:21

no

Organ Dyn. 2021 Oct 31 January-March; 50(1):100802

utf-8

Organ Dyn

10.1016/j.orgdyn.2020.100802

oa_other

==== Front J Public Econ J Public Econ Journal of Public Economics 0047-2727 0047-2727 Elsevier B.V. S0047-2727(21)00025-6 10.1016/j.jpubeco.2021.104389 104389 Article What motivates non-democratic leadership: Evidence from COVID-19 reopenings in China☆ Fisman Raymond a Lin Hui b Sun Cong c Wang Yongxiang de Zhao Daxuan f⁎ a Economics Department, Boston University, Room 304A, Boston, MA 02215, United States b Finance Department, School of Business, Nanjing University, Anzhong Building, 210093 Nanjing, Jiangsu Province, China c School of Urban and Regional Science, Shanghai University of Finance and Economics, Shanghai 200433, China d School of Management and Economics, The Chinese University of Hong Kong, Shenzhen 518172, China e Shanghai Advanced Institute of Finance, Shanghai Jiaotong University, Shanghai 200030, China f Department of Finance, School of Business, Renmin University of China, Beijing 100872, China ⁎ Corresponding author. 18 3 2021 4 2021 18 3 2021 196 104389104389 23 11 2020 17 2 2021 17 2 2021 © 2021 Elsevier B.V. All rights reserved. 2021 Elsevier B.V. Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. We examine Chinese cities’ COVID-19 reopening plans as a window into governments’ economic and social priorities. We measure reopenings based on official government news announcements, and show that these are predicted by citizen discontent, as captured by Baidu searches for terms such as “unemployment” and “protest” in the prior week. The effects are particularly strong early in the epidemic, indicating a priority on initiating economic recovery as early as possible. These results indicate that even a non-democratic government may respond to citizen concerns, possibly to minimize dissent. Keywords Bureaucratic incentives Unrest Non-democracy China COVID-19 ==== Body pmc1 Introduction One conjectured merit of non-democratic governance is that it allows for decision-making that provides coordinated responses to societal concerns. As observed by Kudumatsu and Besley (2008) command-and-control systems may be effective in, for example, providing public infrastructure and coordinating private investments. However, as they also observe, an effective government should also “protect the vulnerable,” which will often require that officials attend to citizens’ concerns. (This point is echoed in a more nuanced form in classic analysis of democratic transitions by Acemoglu and Robinson (2005), which argues that elites may choose to “buy off” some groups (e.g., the middle class) to forestall democratization.) While non-democracies are, by definition, not subject to democratic accountability and hence face no immediate imperative to respond to societal demands, populations that grow too dissatisfied with their leaders may take more extreme measures to bring about regime change, as exemplified by the recent Arab Spring uprisings (among many, many others). Thus, leaders that take the long view may wish to minimize dissent in order to maintain control (see, e.g., King et al., 2013 on the central importance of maintaining social stability for Chinese officials). This imperative for order and stability is almost surely felt also by lower-level officials, as it may be necessary to gain promotion. We study Chinese city leaders’ responses to the recent COVID-19 pandemic. Studying government policies toward the pandemic in a non-democratic setting is worthwhile in its own right, given the vigorous debate on whether democratic accountability has aided or hindered responsiveness to this epochal global health threat (e.g., Kleinfeld, 2020).1 But we may also use these responses as a way of understanding what motivates officials in the Chinese bureaucratic hierarchy more generally. That is, how have governments weighed economic versus public health considerations in guiding their policy responses? And to the extent that they are concerned with economic outcomes, are these primarily citizen concerns, standard governmental performance metrics, or both? Our particular setting is well-suited to studying these questions: the rapidly-shifting nature of the epidemic, combined with possible citizens discontent over extended lockdowns, forced leaders in each city to manage potentially conflicting objectives on a day-to-day basis (and thus generating within-city variation that we exploit in our analysis). We measure government reopenings based on the number of articles in the government-owned city newspaper that includes the term reopening in the title, during the first two months after the Chinese New Year.2 This measure has the advantage of reflecting government policy toward restarting the economy, rather than the combined effects of government policy and individuals’ responses to health risks, such as traffic patterns (which we will also consider).As we explain in greater detail in Section 2.1, the government did not provide explicit reopening policy announcements, as was common in many states and cities within the U.S. Rather, reopening “guidances” came gradually and sometimes obliquely via stories in official newspapers that described what was expected to unfold in subsequent days or weeks. We begin by showing that cities with higher COVID case counts experienced a greater decline in economic activity, as captured by traffic in 2020 versus the same day in 2019 (relative to the Chinese Spring Festival).3 Specifically, in a panel setting, we show that the presence of new COVID cases in the prior week is a negative predictor of traffic. This is a basic reality check that combines both governmental and individual responses. But these public health measures do not correlate with stories in official government newspapers that focus on reopening.4 Government reopening announcements do, however, have a direct positive relationship with economic activity, as captured by traffic flow – more reopening news in the prior week is associated with higher traffic flow, indicating that government pronouncements likely have an impact on real economic activity, distinct from individual concerns resulting from the existence of recent COVID cases. As a proxy for government responsiveness to citizen concerns, we use searches of keywords that are related to social unrest and economic difficulties (such as bankrutpcy, default on mortgage, protests, rights protection; see Section 2.3 for more details) on Baidu, a Chinese internet search platform.5 We show that newspapers have more stories that headline reopening if city residents search more for these terms in the preceding week. This is true for specifications that include city and day fixed effects, so that the link between Baidu searches and reopening announcements is identified using within-city variation in expressions of economic anxieties and frustration. Finally, the relationship is driven by reopening announcements that occur relatively early in the sample period, suggesting that citizen discontent prompted leaders to prioritize economic recovery as early as possible. Our findings contribute to the already-vast literature on the political economy of COVID responses globally. Much of this work has emphasized partisan differences across U.S. states (e.g., Adolph et al., 2020, Allcott et al., 2020, Barrios and Hochberg, 2020, to name just a few). To our knowledge we are the first to study COVID responses in a non-democratic system. We see our main contribution as using these events to understand more broadly the attentiveness of officials to citizen concerns in China, and in non-democratic systems more generally. Particularly in recent years, political economy scholars have looked more rigorously at the incentives and objectives of Chinese officials, documenting, for example, the link between promotion and local economic growth (Jia et al., 2015), pollution (Kahn et al., 2015), and workplace safety (Fisman and Wang, 2017). Most directly related, a small number of recent papers have shown a link between the direct expression of political concern and bureaucratic responses: Our findings complement those of Chen et al. (2016), who show in an audit study that Chinese city leaders are more likely to reply to requests for an unconditional cash transfer program (Dibao)when the letter includes threat of collective action. Our findings indicate that this responsiveness extends to substantial policy actions (whereas in Chen et al., 2016 the outcome is only whether an official provides a written response). Our work contributes to the nascent literature that aims to better understand when and how the Chinese bureaucracy responds to such concerns; most notably Jiang and Zeng (2020) show that responsiveness to citizen petitions on land reform is contingent on city leaders’ connections to higher-level officials, which may allow them to overcome potentially powerful local real estate business interests. More broadly, our work speaks to the objectives of leaders in non-democracies. As we noted at the outset, a range of models suggest that non-democratic leaders may attend to the preferences of their citizens, even in the absence of direct accountability. Our findings do suggest that in the case of China, leaders heed the expressed concerns of their citizens. (Note that this is by no means an endorsement of the system overall – in the model of Acemoglu and Robinson (2005), for example, elites may cater to the concerns of others to retain power, because the alternative – revolution and democracy – is worse for elites, even if it improves the well-being of the median citizen.) As such we add to the literature in political science and economics which considers how non-democratic leaders maintain control. For earlier contributions see, especially, Svolik, 2012 and citations therein; and for recent related work on China specifically, see Li (2014) on labor disputes and public goods provision, and Distelhorst and Hou (2017) for a field experiment on local government responsiveness to citizens’ complaints. 2 Background and data We draw on a range of distinct data sources to conduct our main empirical tests. In the subsections that follow we provide overviews of each data component. 2.1 News reports on COVID-19 reopenings To measure the extent of reopening we utilize news reports via official “mouthpiece” newspapers in each city. Before proceeding to describe our measure, we clarify a few aspects of reopening plans and their communication to the general public. Most importantly, reopening was a local decision – the central government specified the earliest possible reopening date, which applied uniformly across the country, but beyond this constraint, the decision was decentralized to each province, which in turn delegated the decision to each city.6 Within each city, the government rarely provided explicit pronouncements, perhaps to avoid taking responsibility for new cases that emerged in the wake of reopenings.7 Rather, city leaders provided policy guidance via official media announcements as described below, and also broadcast guidelines simultaneously via social media outlets like WeChat (See Qin et al. (2018) for a reference). Thus, our best proxy for reopening plans come via these official media sources, whose communications signal the local government’s intentions and desires. Because cities opened district by district and sector by sector, and directives varied in their forcefulness, we cannot generate a simple event study type of analysis with a single reopening date. (Some sample news stories may help to illustrate. We provide some examples from the Jiefang Daily, Shanghai’s official newspaper. Reopening reports in our dataset include, among others, the following, which illustrate the geographic and sectoral specificity of announcements: “93 percent of regional headquarters of international companies in Shanghai reopened” (February 23, 2020); “Top industrial 100 firms in Pudong mostly reopened: (February 25, 2020); “Complete reopening of household management service still needs to wait” (February 23, 2020); “More than 96 percent of large industrial firms reopened” (February 28, 2020); “More than 91.7 percent of firms affiliated with central-government-owned enterprises restarted” (March 2, 2020); “Some the parks in Shanghai gradually reopening” (March 12, 2020); “Key project recovery reaches 86 percent in Shanghai” (March 16, 2020); “Restart of cultural and tourism industries in Shanghai accelerated” (March 20, 2020).) In China, the media is monopolized and tightly controlled by the state. Each level of government and its corresponding CPC leadership generally owns one official newspaper as its “mouthpiece” to deliver local political information, propagate official policy, and guide public opinions (Qin et al., 2018).8 The news reported in these local papers thus represent the views and policies of the local government and its CPC counterpart, rather than those of individual citizens. Our starting sample of news stories includes the titles of 452,965 articles published between February 10 and April 15, 2020 from 216 local official newspapers, belonging to local government entities for 212 locally-controlled prefectural cities and the 4 municipalities that are directly controlled by the central government (Beijing, Chongqing, Shanghai and Tianjin).9 As noted earlier, we use February 10 as the start date because it is the first business day after the Chinese Spring Festival; we ended our data collection on April 15. We first identify whether these news reports are related to the recovery of the economy after the epidemic of COVID-19 based on a matching of “reopening” keywords. These are terms that relate to the recovery of firms, recovery of schools, and recovery of the market.10 Stories with titles that include the name or names of other countries are excluded since these reports generally refer to news of reopenings external to the prefecture. In addition, a title with negative words is also excluded.11 We count the number of reopening-related reports for each day in each city and use it to measure the extent of local reopening and recovery as proclaimed by the local government. The main variable we use in our analysis is log(1+Reopening) because of the long right tail in reopening announcement frequency. (In practice our results are qualitatively the same in terms of implied magnitudes and statistical significance if we use Reopening or the inverse hyperbolic sine transformation.) 2.2 Within-city traffic flow We obtain traffic flow data from Baidu Qianxi (https://qianxi.baidu.com/), which constructs various daily indices of population movements across and within cities, by tracking geographical coordinates of mobile phone users (also see Chen et al. (2020)for a reference). We use the Baidu Qianxi index for within-city traffic flow, which captures daily human movement within a city. Baidu Qianxi defines the index as the number of people who left their homes on a given day as a fraction of overall city population. To capture the influence of COVID-19 on city traffic flow, we calculate abnormal daily traffic by dividing the daily traffic index for each day during February 10 – April 15, 2020, by the corresponding daily traffic index in 2019. To generate this excess traffic measure (ΔTraffic) – as with other “excess activity” metrics we describe below – we match Feb 10, 2020 to Feb 11, 2019. These dates are both Mondays, and both are the first working day after the Spring Festival, when economic activity in China typically returns to normal after the Chinese New Year holidays. This normalization is central, since traffic indexes are entirely non-comparable across prefectures. For example, on most dates throughout 2019–2020, Beijing’s traffic index is lower than Bayannur, a virtually unknown municipality in Inner Mongolia. 2.3 Measuring social unrest Our measure of socia unrest comes from Baidu Index (http://index.baidu.com/), which is similar to Google Trends and is widely used in academia (see Liu et al. (2016) and Li et al. (2019) among many others for references). Baidu records daily searching behavior of all users, and provides search indices disaggregated to the city-level. According to public information provided by Baidu, its search index is based on search volume in a given period, standarized by the past 30-day historical average value of the searching volume of the same key words. 12 We may thus capture the prevalence of certain keyword searches at the city-date level, which proxy for the attention paid by city residents to particular topics at a given point in time.We calculate the average number of searches for a group of keywords related to social unrest and economic difficulties on each day for each city in the preceding seven day period (i.e., our February 10 measure would capture keyword searches during February 3 – 9). These keywords include default (WeiYue), strikes (BaGong), rights protection (WeiQuan), bankruptcy (PoChan), mortgage default (DuanGong), protest (KangYi), unemployment (ShiYe), stagnation (TingZhi), deficit (KuiSun), and going down/declining (XiaHua). We further distinguish these keyword into those that are more politics-related (strikes, rights protection, and protest) versus economic-related (default, bankruptcy, mortgage default, unemployment, stagnation, deficit, and going down/declining) to generate two distinct measures of discontent. The resultant variable is BaiduDissent_7Day; we also report results that use a two-week window (BaiduDissent_14Day). In closing this subsection, we note that our measure of dissent based on Baidu searches is highly correlated with historical collective actions. The China Labor Bulletin (https://clb.org.hk/) has recorded strikes and worker protests in China since 2011, which we use to calculate the number of such collective actions during 2019, and also averaged during 2017–2019. In Appendix Table A2, we show that this historical measure of unrest is correlated with our Baidu-based measures over our sample period, after controlling population and GDP per capita. 2.4 Daily city-level COVID-19 cases The daily case count of COVID-19 in Chinese cities is collected from the Chinese Center for Disease Control and Prevention (http://2019ncov.chinacdc.cn/2019-nCoV/). It reports the confirmed cumulative, dead, and recovered COVID-19 cases in each city each day. We do not include any cities in Hubei province in our sample, as these numbers experienced major adjustments offically after the Wuhan lockdown. Much more important, the reopening of Hubei province was managed by the central government rather than any policies initiated by the local governments, with the Chinese vice Premier Sun Chunlan dispatched to Wuhan in late January to manage Hubei’s coronavirus response; Sun Chunlan was not approved by the CPC Central Committee to return to Beijing until April 27, 2020. 13 The main variable we use in our panel analysis is an indicator variable for any new reported COVID-19 infections in the prior 7 day period (AnyCOVID); the log of (one plus) the number of recent COVID-19 infections yields very similar results. 2.5 Air quality We obtain air quality data from the China National Environmental Monitoring Center, which operates over 1600 monitoring stations throughout China and provides a daily air quality index (AQI) for each city.14 To capture the impact of COVID-19, we again deflate daily AQI in 2020 by corresponding daily AQI in 2019.15 Because of the very long tails in the AQI distribution, we use the log value; the resultant variable has a roughly normal distribution (log(ΔAQI)). We provide summary statistics for city-date variables in Appendix Table 1 . Note that since we exploit both cross-sectional and time variation, we do not require time-invariant city controls. The data reflect observations from 216 prefectures, out a total of 292 in China. As noted earlier, we exclude the observations from Hubei province, and also those lacking newspaper records or population data. Collectively, our 216 prefectures capture 82.9 percent of China’s population and 84.1 percent of the country’s GDP. 16 In Appendix Figure A1, we provide a map of China, labeling the cities that are in our sample, and also those that we do not include. The one notable pattern is that most cities in far north province of Heilongjiang are missing from the sample. Otherwise, our sample is well-distributed throughout the country.Table 1 Reopening announcements, COVID-19 infections, and proxies for economic activity. (1) (2) (3) (4) (5) (6) (7) Dependent Variable ΔTraffic log(ΔAQI) log(1+Reopening) AnyCOVID −0.052∗∗∗ −0.052∗∗∗ −0.122∗∗∗ −0.121∗∗∗ −0.011 (0.008) (0.008) (0.031) (0.031) (0.033) log(1+Reopening) 0.012∗∗ 0.012∗∗ 0.019∗ 0.018∗ (0.005) (0.005) (0.010) (0.010) City FEs Yes Yes Yes Yes Yes Yes Yes Date FEs Yes Yes Yes Yes Yes Yes Yes Observations 7342 7342 7342 9606 9606 9606 9606 R-Squared 0.896 0.894 0.897 0.229 0.226 0.230 0.342 Notes: Numbers in parentheses are robust standard errors. Standard errors are clustered at the city level. The sample covers the period from Feb 10, 2020 to April 15, 2020. The dependent variable in Columns (1)--(3) is ΔTraffic, the daily Baidu traffic index deflated by the corresponding daily traffic index in 2019. The dependent variable in Columns (4)--(6) is log(ΔAQI), the log of the daily AQI in 2020 deflated by the corresponding daily AQI in 2019. The dependent variable in Column (7) is log(1+Reopening), the natural log of one plus the number of reopening-related reports for each day in each city. AnyCOVIDis an indicator variable which denotes that there has been at least one new coronavirus case in each city 7 days preceeding each day. Significance: * significant at 10%; ** significant at 5%; *** significant at 1%. 3 Results 3.1 Preliminaries: COVID cases, reopening plans, and economic activity We begin by exploring whether our measure of government reopening is correlated with standard measures of economic activity, in particular ΔTraffic (Baidu traffic index relative to 2019) and log(ΔAQI) (air quality index relative to 2019).We emphasize that in this section we do not necessarily ascribe any strong causal interpretation to the patterns we document – rather, our goal is to examine whether reopening announcements are correlated with observables (traffic and pollution) which reflect real economic activity. To account for risk from COVID exposure, which may affect economic activity regardless of government reopening plans, we also include as a covariate AnyCOVID, an indicator variable which denotes that there has been at least one new coronavirus case in city c in the 7 days preceding t. Thus, our specifications take the following form:(1) ΔTrafficct=β1∗log(1+Reopening)ct+β2∗AnyCOVIDct+υt+γc+∊ct These specifications include both date (υt) and city (γc) fixed effects, so that the relationship between reopening announcementsand economic activity comes from variation across cities at a particular point in time, and across time within a particular city (recall also that both outcomes are differenced relative to the equivalent figure in 2019). In all cases, standard errors are clustered at the city level. These results appear in Table 1, columns (1)--(3) for ΔTraffic and columns (4)--(6) for log(ΔAQI). Starting with ΔTraffic in column (1), we present a specification with AnyCOVIDas the main covariate; its coefficient is negative and highly significant (p<0.001) and large in magnitude, implying that travel is depressed in the wake of recent COVID cases. In column (2) we look at the relationship between log(1+Reopening) and ΔTraffic; its coefficient is positive and significant at the 5 percent level. When both are included in column (3) the coefficients and their significance are virtually unchanged, suggesting that COVID risk and government reopening announcements exert largely independent effectson city residents’ activities. To provide a sense of the magnitude of these effects, the within-city average standard deviation of 0.05 for ΔTraffic, so the point estimate of -0.05 on AnyCOVIDimplies that a case in the prior week is associated with a one standard deviation reduction in driving. (The patterns are qualitatively the same (both in terms of implied effect size and statistical significance) if we use the number of cases rather than AnyCovid. Our preference is to use a measure which captures the existence of coronavirus since even a single case would trigger a response given its exponential growth.) The implied effect for reopening is more modest – the within-city standard deviation of log(1+Reopening) is 0.6, so that a one standard deviation increase in reopening announcements is associated with just over a 10 percent of a standard deviation increase in traffic. The results on pollution are comparable in implied magnitude (the within-city standard deviation of log(ΔAQI) is about 0.11), though less precisely estimated. Finally, in column (7) we look at whether COVID cases in the prior week predict reopening announcements. Interestingly, they do not – this may be because of the difficulty in pinning down the timing that links prior cases to reopening announcements. It may also be that by February, the appearance of individual cases were not seen by individual municipalities as representing substantive public health threats (though the link from new cases to traffic would suggest that the broader population did react with concern). On an anecdotal level, we observed an immediate and complete shutdown of Beijing following the appearance of a larger COVID-19 cluster in early June, 2020, indicating that wider health threats were seen as reason for a strong governmental response. We conclude from this opening set of results that there is very clear evidence that coronavirus cases reduced economic activity within a city, but did not themselves affect reopening announcements, and evidence that, independent of COVID cases, government reopening plans are correlated with local economic activity. Given the absence of any obvious link from coronavirus cases to reopening plans, it is natural to then ask whether other considerations, perhaps distinct from public health concerns, affected governments’ decisions to restart their economies. 3.2 Citizen discontent and reopening plans We begin by looking at daily data on reopening plans as a function of recent internet searches for words that express discontent, either economic or political. Our main specification is similar to Eq. (1), with log(1+BaiduDissent_7days), as the explanatory variable, which reflects the average number of searches of a group of keywords that are related to social unrest and economic difficulties on each day for each city in the preceding seven day period, and log(1+Reopeningct) (total counts of newspaper titles containing reopening-related keywords) as the outcome:(2) log(1+Reopeningct)=β∗log(1+BaiduDissent_7daysct)+υt+γc+∊ct We present these results in Table 2 . The point estimate on log(1+BaiduDissent_7Days) in column (1) is 0.029 (p<0.01). If we think about this as a semi-elasticity, the implied effect is that a doubling of BaiduDissent will lead to a three percent increase in reopening announcements, a small but discernable impact. In column (2) we allow for a longer, two week window for BaiduDissent, which generates a similar point estimate. Column (3) distinguishes between political-based and economic-based dissent. Interestingly, both contribute to the overall correlation between dissent and reopening, though neither is significant at the 5 percent level, perhaps resulting from great noise resulting from the disaggregation. (Note that the within-city correlation between these two variables is very close to zero, so they contribute independent variation to the within-city results.) In Appendix Table A3, we present a placebo regression in which we predict reopenings with our dissent measures 7 and 14 days in the future; we find no significant relationship or even any consistent sign.Table 2 Baidu keywords related to dissent and city reopenings. (1) (2) (3) Dependent Variable log(1+BaiduDissent_7days) 0.029∗∗∗ (0.011) log(1+BaiduDissent_14days) 0.050∗∗∗ (0.019) log(1+BaiduPolitics_7days) 0.012 (0.009) log(1+BaiduEconomy_7days) 0.017∗ (0.009) City FEs Yes Yes Yes Date FEs Yes Yes Yes Observations 9562 9562 9562 R-Squared 0.345 0.345 0.344 Notes: Numbers in parentheses are standard errors clustered by city. The sample covers the period from Feb 10, 2020 to April 15, 2020. All columns use city-date level observations. The dependent variable in all columns is log(1+Reopening), which captures the number of reopening-related reports for each day in each city. BaiduDissent_7daysand BaiduDissent_14daysare the average number of searches of keywords related to social unrest and economic difficulties in the preceding 7 and 14 day period respectively. BaiduPolitics_7daysand BaiduEconomy_7daysare the average number of searches of keywords related to social unrest and economic difficulties, respectively, in the preceding 7 day period. Significance: * significant at 10%; ** significant at 5%; *** significant at 1%. We look at the particular timing of the link between discontent-related searches and reopening in Fig. 1 , which provides a week-specific coefficient for each 5-day work week following February 10 (i.e., the variable BaiduDissent_7days is separated into 9 distinct variables that reflect Baidu searches in each of the 9 weeks following February 10, and zero otherwise). Interestingly, the overall positive relationship is driven entirely by reopening announcements in the first part of the sample, which by definition reflect earlier reopenings. Indeed, beyond the first month the relationship is zero or even negative, perhaps reflecting a substitution to earlier openings for locales that anticipate discontent.Fig. 1 The timing of the relationship between dissent-related keyword searches and reopening announcements. Notes: This figure shows the coefficients from a regression that relates week-by-week Baidu keyword searches to government newspaper reopening articles. Each point indicates the point estimate from a variant on Eq. (2) that allows the coefficient to vary by week. The whiskers show the 95 percent confidence interval of each coefficient estimate. See text for additional details. 4 Conclusion We document a link between citizen discontent as captured via online searches and prefecture reopening announcements, following economic shutdowns caused by the COVID-19 epidemic in China. We suggest that these findings help to shed light on the incentives and motivations of Chinese bureaucrats more generally, supporting the view that officials are motivated to defuse social unrest. We see the main takeaway from our work as showing that incentives for non-elected officials in China are such that they still attend to citizens’ concerns, thus providing a measure of direct acocuntability (in addition to accountability to higher-level officials who decide on promotions). Of course, social unrest itself may impede promotion, given higher-level priorities, but even if this is the underlying explanation for our results, it still leads to local responsiveness to citizen concerns. Turning to the specific setting of COVID-19, our results emphasize the pressures that governments face in balancing public health and economic concerns. As we write in November, 2020, many developing (and some developed) economies have continued their reopening plans despite rising case counts. Our findings indicate that unless leaders find ways of addressing individuals’ economic concerns, citizens may have limited patience for strict public health measures. Appendix A Table A1, Table A2, Table A3 Table A1 Summary Statistics, City-Date level variables. Variable Name Mean StdDev Obs log(1+Reopening) 1.340 0.664 9606 AnyCOVID 0.228 0.420 9606 log(ΔAQI) −0.130 0.535 9606 ΔTraffic 0.830 0.234 7343 log(1+BaiduDissent_7days) 3.954 1.526 9562 log(1+BaiduDissent_14days) 3.981 1.382 9562 Notes: Reopeningis the number of reopening-related reports for each day in each city. AnyCOVIDis an indicator variable which denotes that there has been at least one new coronavirus case in each city 7 days preceeding each day. log(ΔAQI)is daily AQI in 2020 deflated by corresponding daily AQI in 2019. ΔTrafficis the daily Baidu traffic index deflated by corresponding daily traffic index in 2019. BaiduDissent_7daysis the average number of searches of a group of keywords that are related to social unrest and economic difficulties on each day for each city in the preceding 7 day period. BaiduDissent_14daysis the average number of searches of a group of keywords that are related to social unrest and economic difficulties on each day for each city in the preceding 14 day period. Table A2 Relationship between past worker protests and Baidu keywords related to discontent (1) (2) (3) Dependent Variable log(1+BaiduDissent_Total) log(1+Unrest_1yr) 0.412∗∗∗ (0.085) log(1+Unrest_3yrs) 0.551∗∗∗ 0.634∗∗∗ (0.130) (0.133) log(Population) 1.646∗∗∗ 1.533∗∗∗ 1.654∗∗∗ (0.108) (0.140) (0.156) log(GDPPC) 0.823∗∗∗ 0.795∗∗∗ 0.710∗∗∗ (0.066) (0.073) (0.087) Province FEs No No Yes Observations 212 212 211 R-Squared 0.814 0.814 0.885 Notes: Numbers in parentheses are robust standard errors. The sample covers the period from Feb 10, 2020 to April 15, 2020. All columns use city-level observations. The dependent variable in all columns is log(1+BaiduDissent_Total), which captures the log of the total number of searches of keywords related to social unrest and economic difficulties during our sample period. Unrest_1yris the number of recorded strikes and worker protests during 2019. Unrest_3yrs is the average number of recorded strikes and worker protests during 2017–19. Significance: * significant at 10%; ** significant at 5%; *** significant at 1%. Table A3 Baidu keywords related to future dissent and city reopenings. (1) (2) (3) Dependent Variable log(1+Reopening) log(1+BaiduDissent_7fwd) −0.012 (0.010) log(1+BaiduDissent_14fwd) −0.015 (0.022) log(1+BaiduPolitics_7fwd) 0.003 (0.008) log(1+BaiduEconomy_7fwd) −0.010 (0.009) City FEs Yes Yes Yes Date FEs Yes Yes Yes Observations 9562 9562 9562 R-Squared 0.344 0.344 0.344 Notes: Numbers in parentheses are standard errors clustered by city. The sample covers the period from Feb 10, 2020 to April 15, 2020. All columns use city-date level observations. The dependent variable in all columns is log(1+Reopening), which captures the number of reopening-related reports for each day in each city. BaiduDissent_7fwdand BaiduDissent_14fwdare the average number of searches of keywords related to social unrest and economic difficulties in the preceding 7 and 14 day period respectively. BaiduPolitics_7fwdand BaiduEconomy_7fwdare the average number of searches of keywords related to social unrest and economic difficulties, respectively, in the preceding 7 day period. Significance: * significant at 10%; ** significant at 5%; *** significant at 1%. Fig. A1 Fig. A1 Map of cities in the sample. ☆ Fisman and Wang would like to thank the National Science Foundation (grant numbers 1729806 and 1729784 respectively) for financial support. We thank Rui Dong, Beibei Hou, Xiaojia Zheng and Xiaoting Sun for their excellent RA work. Authorship is alphabetical – all authors contributed equally to this manuscript. 1 The Chinese political system is organized around the Chinese Communist Party, which refers to its system of governance as democratic centralism. At lower-levels of government, there may be active citizen input and indeed even multi-candidate elections, though at higher levels in practice political selection is via bureaucratic promotion. See, e.g., Fisman et al. (2020) for additional details. For simplicity we use the term “non-democratic” throughout. 2 As we explain below, we use the last day of the Chinese Spring Festival as Day 0 because there is normally a lull in commerce leading up to this date. 3 This comparison also preserves a comparison between the same day-of-the-week in 2020 versus 2019. 4 We can speculate on the reasons for this lack of correlation. Possibly, it results from the difficulty in pinning down the timing that links prior cases to reopening announcements. It may also be that by February, the appearance of individual cases were not seen by individual municipalities as representing substantive public health threats. On an anecdotal level, we observed an immediate and complete shutdown of Beijing following the appearance of a larger COVID-19 cluster in early June, 2020, indicating that wider health threats were seen as reason for a strong governmental response. 5 It is a common perception that the Chinese government does not tolerate any dissent. However, as we explain in greater detail below, smaller-scale protests and criticism of officials are tolerated and even encouraged in China, as in many non-democracies. 6 The State Council issued such a guidance on Feb 8, 2020. See http://paper.people.com.cn/rmrb/html/2020-02/10/nw.D110000renmrb_20200210_4-03.htm for the full text (Accessed on Aug 12, 2020). For example, Jiangxi Province, like almost all other provinces in China apart from Hubei Province, immediately issued a directive on reopening guidance. This directive echoed the State Council’s order. We emphasize two features of these directives. First, both the central government order and the province directive only outline a set of principles and genereal measures without specifying any particular reopening date. Second, the Jiangxi province directive, which was addressed to all prefectures and counties within Jiangxi Province, explicitly stated in its opening paragraph that its instructions were only guidelines, and that each prefecture/county should implement their own policies based on local conditions. See http://www.jiangxi.gov.cn/art/2020/2/10/art_396_1498548.html for the full text, accessed on Aug 12, 2020. 7 For example, Article 13 of Section 4 in the Jiangxi Province directive specifies that the one who is in charge of reopening decisions should be responsible for all related safety measures and also any COVID-19 related safety issues, see http://www.jiangxi.gov.cn/art/2020/2/10/art_396_1498548.html for the full text (accessed on Aug 12, 2020). Shandong Province issued a very similar directive on Feb 11, 2011 (http://www.shandong.gov.cn/art/2020/2/11/art_97902_347668.html, accessed on Aug 12, 2020), and its last section is titled, “Further emphasizing the responsibilities of enterprises and territorial governments” which stated the same principle: whoever makes a reopening decision – whether local government or business entity – should take responsibility for new COVID-19 cases. 8 China has a parallel power structure, with representation at each level of the Communist Party of China (CPC) – which is largely tasked with formulating policy – and the government, which implements policies. For example (and of particular relevance for our application), the top bureaucrats in each prefecture are the city mayor (representing the government) and the municipal party secretary (representing the CPC). 9 Reopening is sometimes mentioned in commercial advertisements in government-controlled newspapers. These are not included in our sample, as they reflect private intentions rather than official policy. 10 Specifically, our list of keywords is as follows (rough translations in parentheses): FuGong (workers returning to the plant, or employees returning to their offices), FuChan (workers going back to the plant), FuXue (students returning to school), FuKe (students returning to school); FuShi (markets are reopened); FuYe (business operations resume); and also these combination of the following keywords: HuiFu (recovery)+QiYe (firm); HuiFu (recovery)+Shengchan (production); HuiFu (recovery)+Xuexiao (school); HuiFu (recovery)+ShangKe (back to school); HuiFu (recovery)+KaiFang (reopen); HuiFu (recovery)+YingYe(business operation); HuiFu (recovery)+KaiYe (open business); ChongXin (re-)+KaiFang (open business); ChongXin (re-)+YingYe (business operation); ChongXin (re-)+KaiYe (open business). 11 In our practice, we exclude titles with the following Chinese words: “ZanTing” (suspension),” ZanHuan” (postpone), “Yanchi” (delay), “TuiChi” (delay), “TingZhi” (stop), “YanChang” (extend). 12 See for example, https://zhidao.baidu.com/question/17067070.html for a discussion. Accessed on Aug 12, 2020. 13 See “Efforts to contain the coronavirus outbreak a test of China’s centralized control,” Los Angeles Times, January 27, 2020, among many other sources, for news reporting about the central government takeover. 14 See, e.g., Dong et al. (2019), for details on the AQI index. 15 That is, for Feb 10, 2020, we would deflate by February 11, 2019, since both are the first workday after the Chinese Spring Festival and both are Mondays, and so forth. 16 Note that there are fewer observations for a subset of variables. The Baidu traffic indices are available only through March 27, 2020, so the panel is somewhat shorter for that variable. Additionally, some basic controls are unavailable in city yearbooks: of the 216 cities that serve as our main sample, GDP data are unavailable for three cities; two others have not yet reported GDP growth for Q1 of 2020. Finally, data on government finances are unavailable for some cities: 4 do not have fiscal revenue data for 2018, while debt data for 2019 is missing for 22 cities. ==== Refs References Acemoglu D. Robinson J.A. Economic origins of dictatorship and democracy 2005 Cambridge University Press Adolph, C., Amano, K., Bang-Jensen, B., Fullman, N., Wilkerson, J., 2020. Pandemic politics: Timing state-level social distancing responses to covid-19. medRxiv. Allcott, H., Boxell, L., Conway, J., Gentzkow, M., Thaler, M., Yang, D.Y., 2020. Polarization and public health: Partisan differences in social distancing during the coronavirus pandemic. NBER Working Paper (w26946). Barrios, J.M., Hochberg, Y., 2020. Risk perception through the lens of politics in the time of the covid-19 pandemic. Technical report. National Bureau of Economic Research. Chen J. Pan J. Xu Y. Sources of authoritarian responsiveness: A field experiment in china Am. J. Polit. Sci. 60 2 2016 383 400 Chen S. Yang J. Yang W. Wang C. Bärnighausen T. Covid-19 control in china during mass population movements at new year Lancet 395 10226 2020 764 766 32105609 Distelhorst G. Hou Y. Constituency service under nondemocratic rule: Evidence from china J. Polit. 79 3 2017 1024 1040 Dong R. Fisman R. Wang Y. Xu N. Air pollution, affect, and forecasting bias: Evidence from chinese financial analysts J. Financ. Econ. 2019 Fisman R. Shi J. Wang Y. Wu W. Social ties and the selection of china’s political elite Am. Econ. Rev. 110 6 2020 1752 1781 Fisman R. Wang Y. The distortionary effects of incentives in government: Evidence from china’s death ceiling program Am. Econ. J. Appl. Econ. 9 2 2017 202 218 Jia R. Kudamatsu M. Seim D. Political selection in china: The complementary roles of connections and performance J. Eur. Econ. Assoc. 13 4 2015 631 668 Jiang J. Zeng Y. Countering capture: Elite networks and government responsiveness in china’s land market reform J. Polit. 82 1 2020 13 28 Kahn M.E. Li P. Zhao D. Water pollution progress at borders: the role of changes in china’s political promotion incentives Am. Econ. J. Econ. Policy 7 4 2015 223 242 King G. Pan J. Roberts M.E. How censorship in china allows government criticism but silences collective expression Am. Polit. Sci. Rev. 107 2 2013 326 343 Kleinfeld, R., 2020. Do authoritarian or democratic countries handle pandemics better? Commentary March 31, 2020. Kudumatsu, M., Besley, T., 2008. Making autocracy work. Institutions and economic performance, pp. 452–510. Li K. Liu M. Feng Y. Ning C. Ou W. Sun J. Wei W. Liang H. Shao Y. Using baidu search engine to monitor aids epidemics inform for targeted intervention of hiv/aids in china Sci. Rep. 9 1 2019 1 12 30626917 Li Y. Downward accountability in response to collective actions: the political economy of public goods provision in China Econ. Transit. 22 1 2014 69 103 Liu K. Wang T. Yang Z. Huang X. Milinovich G.J. Lu Y. Jing Q. Xia Y. Zhao Z. Yang Y. Using baidu search index to predict dengue outbreak in china Sci. Rep. 6 2016 38040 27905501 Qin B. Strömberg D. Wu Y. Media bias in China Am. Econ. Rev. 108 9 2018 2442 2476 Svolik M.W. The Politics of Authoritarian Rule 2012 Cambridge University Press

0

PMC9753912

NO-CC CODE

2022-12-16 23:26:22

no

J Public Econ. 2021 Apr 18; 196:104389

utf-8

J Public Econ

10.1016/j.jpubeco.2021.104389

oa_other

==== Front Comput Netw Comput Netw Computer Networks 1389-1286 1872-7069 Elsevier B.V. S1389-1286(22)00552-7 10.1016/j.comnet.2022.109518 109518 Article TransMUSE: Transferable Traffic Prediction in MUlti-Service Edge Networks Xu Luyang abc⁎ Liu Haoyu b Song Junping a Li Rui d Hu Yahui e Zhou Xu a⁎⁎ Patras Paul b a Computer Network Information Center, Chinese Academy of Sciences, Building No. 2, 4, Zhongguancun Nansijie, Haidian District, Beijing, 100190, Beijing, China b School of Informatics, The University of Edinburgh, 10 Crichton Street, Edinburgh, EH8 9AB, United Kingdom c University of Chinese Academy of Sciences, No. 19 (A) Yuquan Road, Shijingshan District, Beijing, 100049, China d Samsung AI Center, 50 Station Road, Cambridge, CB1 2JH, United Kingdom e China University of Mining and Technology-Beijing, Ding No. 11 Xueyuan Road, Beijing, 100083, China ⁎ Corresponding author at: Computer Network Information Center, Chinese Academy of Sciences, Building No. 2, 4, Zhongguancun Nansijie, Haidian District, Beijing, 100190, Beijing, China. ⁎⁎ Corresponding author. 11 12 2022 2 2023 11 12 2022 221 109518109518 10 1 2022 28 10 2022 7 12 2022 © 2022 Elsevier B.V. All rights reserved. 2022 Elsevier B.V. Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. The Covid-19 pandemic has forced the workforce to switch to working from home, which has put significant burdens on the management of broadband networks and called for intelligent service-by-service resource optimization at the network edge. In this context, network traffic prediction is crucial for operators to provide reliable connectivity across large geographic regions. Although recent advances in neural network design have demonstrated potential to effectively tackle forecasting, in this work we reveal based on real-world measurements that network traffic across different regions differs widely. As a result, models trained on historical traffic data observed in one region can hardly serve in making accurate predictions in other areas. Training bespoke models for different regions is tempting, but that approach bears significant measurement overhead, is computationally expensive, and does not scale. Therefore, in this paper we propose TransMUSE (Transferable Traffic Prediction in MUlti-Service Edge Networks), a novel deep learning framework that clusters similar services, groups edge-nodes into cohorts by traffic feature similarity, and employs a Transformer-based Multi-service Traffic Prediction Network (TMTPN), which can be directly transferred within a cohort without any customization. We demonstrate that TransMUSE exhibits imperceptible performance degradation in terms of mean absolute error (MAE) when forecasting traffic, compared with settings where a model is trained for each individual edge node. Moreover, our proposed TMTPN architecture outperforms the state-of-the-art, achieving up to 43.21% lower MAE in the multi-service traffic prediction task. To the best of our knowledge, this is the first work that jointly employs model transfer and multi-service traffic prediction to reduce measurement overhead, while providing fine-grained accurate demand forecasts for edge services provisioning. Keywords Edge model transfer Multi-service traffic prediction Service clustering ==== Body pmc1 Introduction Edge computing pushes computation and data storage closer to the user, thereby improving response times and saving communication bandwidth, while serving multiple applications simultaneously, e.g., video streaming, gaming, content delivery, etc. As people work increasingly more often remotely following the Covid-19 outbreak and require network support for different services, the edge computing paradigm is witnessing growing uptake. In order to optimize user experience and operational costs, infrastructure providers have been pursuing dynamic provisioning of network resources based on predictions of user demand [1]. Previous efforts in tackling network traffic prediction frequently exploit the ability of deep neural networks (DNNs) to learn complex patterns from historical data [2], [3], [4], [5], [6], [7]. However, existing solutions either require training one dedicated model for each geographic region and hence have limited transferability (which is of paramount importance in reducing computational costs and the environmental footprint of training DNNs) [2], [3], [4], [5], or disregard essential correlations among services [6], [7]. In practical large-scale network deployments (i) per-service patterns are often distinct within a region, as exemplified in Fig. 1, while (ii) certain areas may exhibit similar characteristics that would allow for direct transfer of models among them, without the need of retraining. These key observations are confirmed by our analysis of a real-world network traffic dataset collected in a major city in Sichuan province, China, serving 2.6 million users, spanning 6.3 square kilometers, and comprising eight edge nodes. This motivates us to propose TransMUSE, a transferable traffic prediction framework in multi-service edge networks, which first groups edge-nodes according to per-service statistical features. Within each cohort, reference neural models are chosen and trained on data collected only in the region with the highest overall traffic consumption, which can be then transferred to other group members. As reference model, we put forward a Transformer-based [8] Multi-service Traffic Prediction Network (TMTPN). Furthermore, we propose WK-means, a service clustering algorithm based on Wasserstein distance to categorize services according to their similarity. We train separately a TMTPN model for each service cluster to boost prediction performance at a regional level. Finally, the reference models are transferred to other regions directly, without adaptation.Fig. 1 A snapshot of the volume of traffic consumed by four different services as observed at an edge node in a network deployment in Sichuan (China) over one week. Our proposed model transfer framework, TransMUSE, provides a comprehensive and cost-effective solution for traffic prediction in multi-service edge networks. The key advantages of TransMUSE are as follows: (i) it provides a model transfer approach among edge nodes to reduce measurement and computational overhead without compromising prediction accuracy – compared with training a model individually on local data for each edge node, TransMUSE exhibits imperceptible performance degradation, with only 1.7% and 0.26% higher MAE and RMSE, respectively; (ii) the proposed TMTPN takes service correlation into consideration to further reduce overhead and the energy that would have otherwise been required to maintain a separate prediction model for each service; our experiments demonstrate that TMTPN outperforms the state-of-the-art MTNet benchmark [2] on the multi-service traffic prediction task by 18.74% and 18.49%, in terms of MAE and respectively RMSE; (iii) the WK-means service clustering tackles both model under-fitting and speed of convergence, improving the TMTPN prediction performance, as it attains 17.59% and 27.89% lower MAE and respectively RMSE, as compared to predicting without prior service clustering. To the best of our knowledge, TransMUSE is the first multi-service traffic forecasting solution for edge networks that leverages model transfer and service clustering to achieve high accuracy at a low measurement cost. The rest of the paper is organized as follows. The multi-service prediction problem is formalized in Section 2. The proposed TransMUSE framework is discussed in detail in Section 3. Section 4 provides exhaustive experimental results to demonstrate TransMUSE’s efficacy. Section 5 discusses the most relevant related work and Section 6 concludes the paper. 2 Problem formulation Our aim is to address the challenges of handling spatial heterogeneity of service traffic in edge networks and reducing model training costs when forecasting future demands in edge networks, to support the effective management of their resources. Formally, multi-service traffic forecasting seeks to maximize the probability that, given T previous measurements of the traffic volume consumed by K services, the predicted traffic consumption over F future time steps is as close as possible to the ground truth. Denoting by xtk the traffic volume of service k at timestamp t and Xt≔[xt1,…,xtK] the snapshot of all K services at time t, and considering a forecasting model that is parameterized by θ, the multi-service traffic forecasting problem is equivalent to: arg maxθpθXt+1,…,Xt+F|Xt−T+1,…,Xt. To solve this problem, we design a Transformer-based Multi-service Traffic Prediction Network (TMTPN) that captures temporal correlations among traffic time series via multi-head attention, then improve forecasting accuracy via service-clustering, as we detail next. Fig. 2 The proposed TransMUSE framework incorporates five stages: (1) Edge nodes are grouped into several node clusters, within which model transfer is to be conducted; (2) In each cluster, the edge node with the largest traffic volume is selected as reference (highlighted with hashed patterns); (3) At each reference node, services are further partitioned into service clusters by WK-Means; (4) One TMPTN is trained for each service cluster; (5) The models trained on reference nodes are transferred to the recipients (highlighted on the right) within the corresponding node clusters. 3 Transmuse framework We propose TransMUSE, a deep learning framework for accurate and cost-effective multi-service forecasting at the network edge. Fig. 2 gives an overview of the different components this framework entails and the relationship between them, namely: 1. Edge Node Clustering: We cluster edge nodes by a set of service-level statistical features using the K-means algorithm, to determine the neural model transfer scope. 2. Reference Node Selection: Within each scope, we select the node with the overall highest traffic consumption as the reference node; reference neural models for forecasting will be trained with data collected at such nodes. 3. Service Grouping: As certain mobile services exhibit statistical similarities, we cluster services using a modified K-means algorithm based on Wasserstein distance, aiming to reduce the number of multi-service neural models to be employed for prediction. 4. Model Training: At the level of each reference node, we train a dedicated TMTPN model for each service cluster, which will simultaneously predict the volume of traffic for all services within such clusters. 5. Model Transfer: We transfer the trained reference models from each reference edge node to all other nodes within the corresponding clusters, where they will be applied for inference without further training. Next, we discuss in details the key stages of our TransMUSE framework. 3.1 Edge model transfer In Multi-access Edge Computing (MEC) scenarios, it is often impractical to train a neural model at each individual edge node, as their computational power is limited and the operational costs and energy expenditure can become prohibitive to operators when deployment density increases. Edge model transfer aims to reduce the cost of measurement collection and model training, by confining these tasks to designated nodes and reusing models trained there on other nodes, without further local tuning. Different from cloud–edge approaches where a central node maintains a global model refined through model updates resulting from local training (federated learning), edge model transfer only considers the model to be transferred among edge nodes without the need for a central cloud. This brings additional merits in terms of data privacy and communication overhead reduction, as the transfer process is confined within a limited scope. A model to be transferred is called a reference model, the node where a reference model is trained is called a reference node, and the edge nodes that adopt it are referred to as recipients. There are two key issues to address in the edge model transfer process. The first, is determining the scope of model transfer. Different edge nodes may observe distinct traffic patterns due to geographic dissimilarities in terms of mobile user demographics [9] or socioeconomic function (residential areas, business districts, shopping centers, etc.). Edge model transfer, therefore, should be applied across edge nodes (within a cluster) with similar traffic features. Secondly, choosing at which edge node to train a reference model to be transferred within the corresponding cluster will impact the inference accuracy. We put forward an edge model transfer strategy that deals with these two issues as follows: • Determining Transfer Scope: We use K-means clustering to group edge nodes according to nine statistical features, i.e., mean, standard deviation, maximum value, minimum value, skewness, kurtosis, and the 1st, 2nd, and 3rd quartiles of traffic volume for each service over the historical traffic data for model training. Specifically, we used an 8:1:1 data split for training, validation and testing. With 20 services, each edge node is represented by a matrix of shape 20 × 9. A model will only be transferred within the same cluster of edge nodes. In the deployment process in a real network environment, the operator will have the freedom to select a broader range of relevant statistical features for this task. • Reference Model Training: Within each cluster, a reference model will be trained only with data from the edge node where the overall highest traffic consumption is observed. Reference models are then transferred to the recipients within the corresponding clusters. The results we present in Section 4 confirm the generalization abilities of this approach. At the level of a reference node, a set of neural models will be trained, each of which targets future traffic predictions for groups of services with similar characteristics, as we explain next. 3.2 WK-means service clustering Traffic patterns and volumes may differ among services due to content popularity, number of service subscribers, service scope, etc. (see Fig. 1), leading to high information entropy if observing all services together. Note however that it would be impractical and expensive to maintain a specific prediction model for each service. Therefore, we propose a service clustering algorithm based on the Wasserstein distance (WD) between per-service time-series data points, which we name WK-means. We resort to WK-means to reduce the number of prediction models by grouping services into clusters according to their traffic pattern similarity. In such manner, we employ a single model for each service cluster, which can learn specific patterns corresponding to the unique cluster features. Accordingly, we do not have to maintain too many models, thereby reducing training cost and potentially saving energy. By grouping services with WK-means, we ensure the model learns from enough data and a rich set of features, so that it does not perform poorly when predicting on the test set (i.e. we avoid under-fitting). There are two key factors to consider when measuring time-series similarity, namely, magnitude and ‘shape’. The former indicates how comparable the traffic volume of different services is; the latter indicates any similarities in terms of periodicity and short-term temporal patterns. The WD takes these two factor into consideration at the same time, which makes it particularly suitable for our grouping task. Originally the WD was proposed to measure the similarity between two probability distributions, and was recently employed in optimal transportation problems [10]: (1) Wp(P,Q)=infμ∈Γ(P,Q)∫ρ(x,y)pdμ(x,y)1/p, where P and Q are two probability distributions in Rd, and Γ(P,Q) is the set of all probability measures on Rd×Rd with marginals P and Q. ρ(x,y)p is a measure of distance between x∈P and y∈Q (e.g., p=2 for Earth mover’s distance). Intuitively, the WD represents the minimal distance for moving the mass of distribution P to exactly fit the mass of distribution Q. Unlike other distance metrics, such as Euclidean distance, Jensen–Shannon (JS) divergence or Kullback–Leibler (KL) divergence, WD has the following key advantages: (i) if the target distributions lie in low-dimensional manifolds or share disjoint support, which is not uncommon for high-dimensional data, WD offers a more informative measure (which is not the case for KL and JS divergence that return a constant value or infinity) [11], [12]; and (ii) WD maintains the underlying geometry of the space [13], that is, it not only takes the quantitative value into consideration, but also pays attention to the similarity of distributions’ shapes. In contrast, the Euclidean distance cannot quantify shape differences or capture the degree of changes between two times series [14]. Based on WD, we propose the WK-means service clustering algorithm, summarized by the pseudo-code in Algorithm 1. To generate N clusters from S services, WK-means initially sorts all the services by their volume and splits the sorted sequence at SN,2SN,…,(N−1)SN. That is, the sorted sequence is evenly divided into N segments (lines 2-6). WK-means further chooses the service in the middle of each segment as cluster center (lines 8-13). Instead of using random initialization, this approach speeds up the convergence process. Then, the WD between each service and sub-cluster center is calculated, and each service is re-assigned to its nearest sub-cluster (lines 14-24). Finally, each sub-cluster center is updated (line 25) and the previous two steps are iterated until all sub-clusters convergence or the iteration epoch reaches a predefined limit (lines 7-26). In a setting with multiple edge nodes, it is possible that WK-means may generate different service clustering results on different nodes. To comply with our model transfer strategy, we first conduct WK-means at the level of every edge-node and select the most frequent clustering pattern as the global service grouping. Such pattern is applied to all other nodes in our design. Fig. 3 TMTPN architecture based on an Encoder–Decoder design. Historical traffic input combined with positional encoding (PE) is processed by the Encoder, which gives the output to each Decoder block. Within each, the core components are a multi-head Attention block and a Linear block. 3.3 TMTPN model design To perform multi-service traffic forecasting, we design Transformer-based Multi-service Traffic Prediction Networks (TMTPNs), each of which inherits from the canonical Transformer architecture and is dedicated to each individual service cluster. Transformers have shown remarkable performance in processing sequential data and have been adopted previously for natural language processing [15], computer vision [16], and vehicular traffic prediction [17] tasks. Our TMTPN model is illustrated in Fig. 3 and follows an Encoder–Decoder paradigm, encompassing the following components: • Multi-Head Attention: Multi-head attention consists of multiple scaled dot-product attention structures that capture temporal dependencies in long sequences. The attention block receives three inputs: Q∈RT×dk (query), K∈RT×dk (key) and V∈RT×dk (value), in which T represents the sequence length and dk is the embedding dimension of each item in the sequence. Attention is computed as: Attention(Q,K,V)=softmaxQKTdkV. QKT generates a T×T matrix of alignment scores, where each entry denotes the correlations between two instances in the sequence. The matrix is scaled and then multiplied by V to generate the hidden representation of the input that incorporates attention information. Multi-head attention splits Q, K and V into multiple chunks, which are processed with independent attention blocks. The outputs of all the attention blocks are concatenated and projected back into hyperspace Rdk. • Encoder and Decoder Layers: The encoder layers (Fig. 3 left) contain a multi-head attention block and a linear block, each of which utilizes a skip connection and layer normalization to prevent over-fitting. For the encoder, only the input sequence is given to the multi-head attention block, i.e., X=Q=K=V, where self-attention is computed. The decoder (Fig. 3 right) incorporates an extra attention block. Specifically, the first attention block in the decoder computes the self-attentional representations of the decoder input, and the second block takes the encoder output as the key K∈RT×dk and the value V∈RT×dk, querying which historical inputs are important when making future predictions. • Positional Encoding (PE): Since transformers do not contain any sequential structure, timing features are not encoded in the network by default. Therefore, positional encoding is added to the input sequence, which reflects the relative position of each timestamp. PE is computed as: PE(pos,2i)=sin(pos/10,0002i/dmodel), PE(pos,2i+1)=cos(pos/10,0002i/dmodel), where pos denotes the position index of the item in the sequence and dmodel is the dimension of the encoded position. • Parallel Decoding: Traditional seq2seq models [7] perform decoding in an auto-regressive manner during training. That is, decoding the tth element in a sequence relies on the hidden states passed from timestamp t−1 and the decoded (t−1)th item, which are provided as the input. It is therefore impossible to decode all the items in parallel. Transformers overcome this problem during training by introducing the shifted decoder input and look-ahead mask. Assume that the ground truth to be provided to the decoder is Y=[y1,…,yF], then the input is the shifted-right ground truth X=[0,y1,…,yF−1]. The look-ahead mask (M) is introduced when computing the alignment scores as follows: A=softmaxQKTdkM, where X=Q=K, and M is a F×F matrix with each entry above the diagonal equal to negative infinity, and below/on the diagonal equal to 0. The scaled matrix of alignment scores is masked with M, which yield a F×F lower triangular matrix, meaning that at a given timestamp i, there is no correlation (Aij=0) with the input from any future timestamp j(j>i). By masking, the decoder can approximate the output at F timestamps, Yˆ=[yˆ1,…,yˆF], in parallel. This technique is only applied during training, while during testing the transformer decodes step-by-step, as seq2seq models. Overall, the proposed TMTPN architecture has several merits: (i) it can be trained fast due as the look-ahead mask and the shifted decoder input that facilitate parallelization; (ii) it can process longer sequences than traditional seq2seq models; and (iii) it captures the most essential historical information that impacts most on prediction results, irrespective of the length of an input sequence, thanks to Position Encoding and Multi-Head Attention. 4 Experiments We implement TransMUSE and its TMTPN models, as well as a set of benchmark neural models in Tensorflow v2.3.0 using the cuDNN v7.6 and CUDA v10.1 libraries. To demonstrate the performance gains of our solution, we train and evaluate the neural models and experiment on a large-scale real-world wired network traffic dataset collected by a network operator in Sichuan Province, China. For this, we employ a high-performance computing cluster comprising 12 servers, each equipped with a 32-cores Intel E5-2620 CPU and running Red Hat Enterprise Linux, and accelerate the training process with multiple GPUs out of a pool of 96 Nvidia RT2080Ti units. We conduct three sets of experiments to demonstrate (1) multi-service traffic prediction performance gains attained by our TMTPN models; (2) the benefit of employing service clustering with WK-means; and (3) forecasting performance with edge model transfer. 4.1 Dataset & pre-processing The dataset we employ was collected in a city with over 6 km2 land coverage, administratively divided into 7 districts and 1 core urban area, and with a population of approximately 2.7 million inhabitants. The traffic within each district (D1 to D8) is handled by a dedicated edge node, and the high level structure of the deployment is illustrated in Fig. 4. Traffic data was collected by Deep Packet Inspection (DPI) via port mirroring, between July 1st and 31st, 2020. Traffic was aggregated at session level, with only application type, district identifier, direction (uplink/downlink), total volume and timing information (session start/end) being recorded, to preserved anonymity. In total, 20 service types are distinguished, as summarized in Table 1, where these are sorted in descending order by their volume across the entire deployment, and indexed. The traffic volume distribution for the top-8 services is shown in Fig. 5, where bars depict the fraction of the overall volume and the line the corresponding values. Due to the commercially sensitive nature of the dataset, we cannot disclose the precise identity of the city, nor the specific service names for which traffic measurements were collected. Over the 31 consecutive days of measurements, we sample the traffic consumption every minute, assuming uniform consumption per session throughout their duration. This is reasonable, given the predominantly short-lived nature of sessions, leading to temporal sequences of 44,640 data points for each service in each region. We normalize service traffic volumes to the 0–1 range, to ensure similar magnitudes during training. We use an 80/10/10 data split for training, validation, and testing, and train models separately on a region-by-region basis.Fig. 4 Network topology and district map of the target city. Traffic measurements collected via DPI and further processed at the core router. Table 1 Service names and indexing for the traffic dataset used in our experiments. 1: Web Video 2: Generic Apps 3: Other Apps 4: P2P VOD 5: Chat 6: P2P Download 7: Online Games 8: P2P Video 9: Cloud Storage 10: Shopping Online 11: Live Stream 12: Music Stream 13: News Apps 14: Generic Video 15: VoIP 16: Stock Apps 17: Traveling Websites 18: Mail Apps 19: Living Apps 20: Portal Webs Fig. 5 Service traffic volume (line) and fraction of the total (bar) in the city across 8 districts over 31 days. 4.2 Benchmarks & metrics For comparison, we consider the following state-of-the-art DL models as baselines: • LSTM, which is now a classic structure for tackling regression tasks, and has been extensively used for traffic prediction [18], [19]. We implement a three-layer LSTM, which offers an appropriate complexity-effectiveness tradeoff. • MTNet, which was designed for multivariate time series prediction and adopts an encoder–decoder architecture to extract both long- and short-term hidden representations correlation among these [2]. • GraphConv, which is also aimed at tackling multiple time series predictions [20], integrating graph convolution [21] and an LSTM network to extract correlations between multiple sequences and temporal patterns. We employ the Spektral library for our implementation [22]. • AttentionAR, a model that we implement based on the Bahdanau Attention structure [23] with rolling prediction through a LSTM cell. Attention is used to assign weights to historical input. • GASTN, which was proposed in [24] for mobile traffic prediction based on attention and recurrent neural networks. To evaluate the performance of our proposed models and that of the benchmarks considered, we compute the Mean Absolute Error (MAE) and Root Mean Square Error (RMSE). In essence, these metrics quantify the difference between ground-truth and predicted values, and are defined as follows [25]: (2) MAE=1S×F∑i=1S∑j=1F|yij−yˆij|; (3) RMSE=1S×F∑i=1S∑j=1F(yij−yˆij)2, where S is the number of services, T represents the number of prediction steps, and yij and yijˆ denote the ground truth and respectively predicted traffic volume for service i at timestamp j. Table 2 MAE and RMSE performance (in MB) on the multi-service traffic forecasting task with LSTM, AttentionAR, GraphConv, MTNet, GASTN and our TMTPN across 8 districts. Model\District D1 D2 D3 D4 D5 D6 D7 D8 MAE RMSE MAE RMSE MAE RMSE MAE RMSE MAE RMSE MAE RMSE MAE RMSE MAE RMSE LSTM 45.20 125.90 47.05 136.75 51.37 145.39 21.20 60.98 21.65 63.74 15.69 44.36 18.78 56.67 50.43 147.15 AttentionAR 61.80 176.89 63.08 184.99 65.40 181.71 24.24 66.58 27.79 81.35 19.06 53.05 23.70 70.21 64.11 180.17 GraphConv 107.42 300.25 110.15 327.21 116.61 317.44 36.44 99.49 40.93 123.34 23.18 66.13 34.41 108.22 118.35 340.68 MTNet 44.15 127.37 46.78 136.54 48.97 138.71 21.86 62.33 23.70 71.22 16.40 48.22 18.54 57.21 46.51 133.26 GASTN 48.28 137.85 47.91 145.03 53.92 156.82 20.98 58.25 23.30 72.51 15.83 44.88 20.47 61.13 51.29 149.57 TMTPN (ours) 37.69 106.19 38.16 118.76 43.27 124.86 16.57 46.80 17.99 53.91 12.04 34.94 14.91 45.98 41.49 117.44 4.3 Multi-service traffic prediction by TMTPN We first examine TMTPN’s performance vis-a-vis that of the benchmarks considered, then investigate the influence that input/output lengths have on this. 4.3.1 Forecasting comparison We first train and test different models for every district separately, using traffic solely observed within each of these. We take input sequences of length 30 (i.e., 30-min historical data) and predict the traffic volume per service over 5 future timestamps. The obtained results are summarized in Table 2, where lower MAE and RMSE values indicate superior prediction performance. Observe that the TMTPN models we propose consistently outperform the state-of-the-art neural networks considered. In particular, when compared with the second best model, LSTM, our TMTPN reduces the MAE and RMSE on average by 18.95% and respectively 17.74%, across the eight districts. This is because the multi-head attention structure adopted by our design allows the model to jointly extract information from different representation sub-spaces at different points in time, giving higher weights to the most significant historical patterns, to enhance prediction performance. The performance of LSTM and MTNet is relatively similar. GASTN’s performance is inferior to that of MTNet in districts with large traffic volumes, but outperforms MTNet in districts where the traffic volume is smaller, i.e., D4, D5, D6. However, AttentionAR largely overestimates traffic demand, indicating that the traditional attention mechanism is not best-suited to multi-service prediction tasks. Finally, GraphConv performs poorly in comparison with our TMTPN and the other two benchmarks we consider. This is likely because no strong spatial relationships between the different services exist in edge network settings. To better appreciate the forecasting performance of our proposed TMTPN model at service level, in Table 3 we summarize that across two districts with dissimilar service usage patterns, namely D1 and D2, while in Fig. 6 we illustrate 5-step prediction instances performed in the two districts across 5-hour windows (busy hours between 15:00 and 20:00 on 29 July, 2020) for 4 randomly selected services (chat, web video, live streaming, P2P video). We only compare TMTPN with LSTM in this and the subsequent experiments, because LSTM appears to be an effective deep learning model that achieves solid performance, which the results in Table 2 and prior work [6], [7] confirm. As can been seen from the figure, TMTPN is superior to LSTM, as it tracks more closely the ground truth traffic that would be available under ideal circumstances. This is especially clear to observe on ‘Chat’ traffic forecasting in D1 (sub-figure (a)) and ‘Live Streaming’ traffic in D2 (sub-figure (b)). Table 3 Per-service prediction performance in terms of MAE (MB) in districts D1 and D2. Service names given in Table 1 by index. DIS Index 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 D1 TMTPN 187.05 250.79 97.91 52.54 23.71 31.94 5.36 10.85 16.71 24.90 4.32 11.01 2.77 1.13 0.59 0.58 0.29 0.12 0.12 0.03 LSTM 257.40 310.18 121.12 65.28 33.13 49.63 7.92 15.47 17.39 36.40 6.26 10.35 2.81 2.00 0.82 0.70 0.26 0.10 0.09 0.03 D2 TMTPN 270.20 241.52 113.04 56.58 29.85 34.36 6.46 10.15 17.50 8.83 5.66 9.30 2.53 1.04 0.53 0.43 0.08 0.07 0.13 0.03 LSTM 277.28 293.57 125.73 68.20 30.70 47.22 8.12 15.04 18.88 11.03 7.09 8.73 2.65 1.49 0.67 0.45 0.08 0.06 0.12 0.03 Fig. 6 5-step forecasting performance with TMTPN and LSTM at service level over 5 busy hours (15:00 to 20:00 on 29 July, 2020) vs Ground truth. 4.3.2 Impact of input and output length In this subsection, we evaluate the long-term and short-term prediction performance of our TMTPN, focusing again on districts D1 and D2. We examine the MAE across all services as the forecasting horizon varies between 5 and 30 steps, while we also vary the input size, i.e., 5, 15 and 30 historical traffic snapshots. The obtained results are summarized in Fig. 7, where the x-axis represents the combination of input and output length (e.g., 30-5 indicates the model uses the previous 30 min traffic data to predict the upcoming 5 min traffic demand). The top sub-figure corresponds to district D1 and the bottom to district D2. Observe that TMTPN is consistently superior to LSTM, as it achieves lower prediction errors. The performance gains grow with the length of the forecasting horizon, with TMTPN reducing the MAE experienced with LSTM on long-term predictions by 43.21% and 40.77% in district D1 and D2, respectively. Benefits are also observable short-term, where TMTPN attains 15.02% and respectively 22.74% lower MAE than LSTM in the two districts, when the input and output lengths are both 5 (5-5). These gains can be attributed to the multi-head attention mechanism that our design adopts. In addition, the shifted input with look-ahead mask not only enables training parallelization, but also ensures TMTPN can predict the future sequence on a rolling basis, unlike the LSTM, which predicts multiple future steps at once and is thus prone to larger errors. Lastly, we note that the input length has only marginal impact on TMTPN’s forecasting accuracy, with input size impacting performance slightly differently at the level of the two districts examined. Yet in both cases the best performance is attained with 15 historical snapshots. Based on these results, we argue that if the input length is too short (5), the model may not be able to capture certain periodic information or longer trends. Fig. 7 Impact of different Input–Output lengths on forecasting performance (in terms of MAE) with LSTM and our TMTPN in districts D1 (top) and D2 (bottom). 4.4 Multi-service clustering Recall that the aim of service clustering in TransMUSE is to further improve forecasting performance by grouping services into different clusters, according to their temporal similarity. Here, we demonstrate the benefits of using our WK-means algorithm for this task (hereafter denoted as WASS), as compared to three benchmarks that can be applied to time series data, namely K-means clustering based on Euclidean distance (EUC), K-means based on Cosine similarity [26] (COS), and (3) Derivative Dynamic Time Warping (DTW) clustering [6] implemented with the tslearn library [27]. Fig. 8 Clustering visualization for the four algorithms. Service bars with the same color are grouped in the same cluster. Service indexes are sorted in descending by traffic volume, and the service name can be obtained by mapping in Table 1. Fig. 9 Service cluster membership and traffic pattern visualization over one week in District No. 2 (D2) when using WASS (left) and DTW (right).. 4.4.1 Number of clusters Before comparing the clustering algorithms, the appropriate number of clusters K needs to be determined. The Silhouette score is routinely employed to characterize clustering performance, which is computed as the difference between the mean of the intra-cluster distances and the mean of the nearest-cluster distances, normalized by the maximum between the two [28]. The silhouette score ranges between [−1, 1], where 1 indicates the best cluster separation, values near 0 indicate overlapping clusters, and negative ones suggest that a sample has been assigned to the wrong cluster. With this, we validate the effectiveness of our WK-means algorithm vis-a-vis that of the benchmarks considered, on the eight districts separately. For each district, K is chosen in the {2,…,5} range, and we compute the silhouette score for each K value. The results on district D1 are given in Table 4, which suggest K=2 is the optimum value. The same holds for the vast majority of other districts, with K=3 yielding marginally higher silhouette scores (0.01 difference) in 2 out of 32 instances. Hence we select K=2 for all the remaining experiments. When the number of services grows, the same approach is suitable. Table 4 Silhouette score comparison with associated distance metric for different numbers of clusters K and the four clustering algorithms considered, in district D1. K\Algorithm EUC COS DTW WASS 2 0.851 0.157 0.852 0.869 3 0.788 0.06 0.814 0.779 4 0.706 −0.09 0.782 0.801 5 0.651 −0.109 0.712 0.732 4.4.2 Clustering algorithms comparison Next we evaluate forecasting performance with TMTPN when a model is trained individually on services clusters, following grouping of the 20 services into K=2 clusters using the proposed WK-means and the benchmark algorithms. We resort again to MAE and RMSE for evaluation and summarize the results obtained in Table 5. The results demonstrate that all clustering algorithms can reduce the prediction errors, which is more apparent in districts with larger traffic volumes, such as D1, D2, D3 and D8. Our WASS solution is superior to COS, because cosine similarity gives priority to the direction of two vectors, such as the semantic similarity between two sentences. DTW and EUC are essentially based on the Euclidean distance between traffic magnitude, whereas the “shape” of a time series is an important feature when measuring the similarity between two time series. Our WK-means algorithm (WASS) based on Wasserstein distance possesses such ability, which is reflected in the lower prediction errors obtained (bottom row in Table 5). Table 5 Clustering algorithms comparison based on MAE and RMSE (MB) of forecasts obtained with TMTPN applied on service clusters across the 8 districts. Algorithm\District D1 D2 D3 D4 D5 D6 D7 D8 MAE RMSE MAE RMSE MAE RMSE MAE RMSE MAE RMSE MAE RMSE MAE RMSE MAE RMSE No Clustering 37.69 106.19 38.16 118.76 43.26 124.86 16.57 46.80 17.99 53.91 12.04 34.94 14.91 45.98 41.49 117.43 EUC 31.22 81.26 31.68 86.91 33.67 88.83 15.00 42.63 16.49 47.62 10.80 30.77 13.21 39.82 33.04 91.57 COS 35.78 100.92 37.18 109.03 41.08 116.10 15.67 43.47 17.83 52.98 11.74 34.19 14.46 43.59 38.65 111.33 DTW 31.22 81.26 33.35 92.38 35.92 98.38 15.45 42.49 16.49 47.62 10.80 30.77 13.21 39.82 34.82 92.63 WASS(ours) 28.98 81.14 30.60 85.09 32.20 72.87 14.65 39.36 15.76 46.91 10.45 30.37 12.99 39.14 32.32 87.95 4.4.3 Cluster visualization To better appreciate where the differences in the performance attained with WK-means and the 3 benchmarks stem from, in Fig. 8, we visualize the cluster membership of the different services in a randomly chosen district (D2). We can draw the conclusion that the traffic magnitude and ‘shape’ do have an impact on the clustering results. Observe that both DTW and EUC rely on the Euclidean distance and only services with very large traffic magnitude are grouped into the same cluster. The fact that service No. 2 belongs to different clusters in DTW and EUC confirms the importance of service traffic shape. In contrast, WASS clusters services from 1 to 6 into one category even though their traffic quantities are distinct. Fig. 9 further illustrates the importance of traffic ‘shape’ as captured with our WASS approach. When we compare cluster 1 and cluster 2 in each sub-figure, it is obvious that the traffic volume range is different. Observe the sub-figures corresponding to WASS, where the services in each cluster share similar traffic patterns, albeit having services with different traffic volume in cluster 1. In contrast, the service shapes in cluster 2 obtained with DTW are more heterogeneous. Cosine similarity pays more attention to the difference between two vectors in direction rather than distance or length. In our service clustering task, the traffic magnitude is the primary consideration, therefore cosine similarity is less effective in clustering service time series, which is also confirmed by our previous results reported in Table 5, where COS performs worst than the other three algorithms in all 8 districts. 4.5 Edge model transfer Finally, we demonstrate the merits of model transfer in TransMUSE by showing that reusing models trained at reference nodes within a node cluster, with the aim of reducing computational overhead, does not impact negatively on the forecasting performance. 4.5.1 Region clustering Recall that the first step in transferring reference models is to decide the transfer scope. We use K-means clustering to group regions according to the statistical features of all the service traffic time series. We resort again to the silhouette score to determine the optimal number of region clusters, which we compute for k∈{2,3,4,5,6,7} in Table 6. We conclude that k=2 produces the highest score and districts D1, D2, D3, D5, and D8 should be grouped together, with the remaining 3 regions belonging to the second edge node cluster. Our focus is on maintaining a small number of models while ensuring model transferability. Our real-world dataset contains district-level edge nodes (rather than base stations). In deployments with many more edge nodes, one can put constraints on the maximum number of groups/the maximum number of members in a group, to maintain a desired complexity/transferability trade-off. Table 6 Silhouette score of edge node clustering by K-means based on 9 statistical features with different number of clusters k from 2 to 7. k 2 3 4 5 6 7 Score 0.441 0.380 0.341 0.259 0.112 0.109 4.5.2 Model transfer validation We order the regions according to the overall traffic volume and conclude that D4 and D3 are to be selected as reference nodes for cluster 1 and 2, respectively. We quantify the generalization ability of the models trained by comparing the RMSE when performing forecasting following model transfer (TransMUSE) versus when models are trained locally at individual region level (Original). To add further perspective and verify our hypothesis that models trained at edge nodes witnessing large traffic volumes have stronger generalization abilities, we also examine the forecasting performance when models are trained on regions where the traffic volume is the lowest among cluster members, prior to transfer (Ctrl-Exp). The result are illustrated in Fig. 10 for the two clusters, where regions appear in descending order by the overall traffic volume. Observe that when the reference models are trained on regions with the highest traffic demand (TransMUSE), the RMSE values are almost identical to those obtained when training models individually at each edge node. The largest performance gap is at the level of D6, where a 0.26% performance degradation is observed in terms of forecasting accuracy (RMSE). In contrast, if reference models were to be trained at edge nodes with lowest traffic volumes (D6 in cluster 1 and D5 in cluster 2), the forecasting performance would suffer (Ctrl-Exp). Specifically, the averaged RMSE error over 8 regions is 9.26 MB, which is 92 times larger than with TransMUSE.Fig. 10 Traffic forecasting performance comparison when model transfer is employed based on highest traffic demand (TransMUSE), a control experiment where reference models are trained at lightly-loaded edge nodes (Ctrl-Exp), and no edge node clustering is performed, i.e. models trained individually at each location (Original). We conclude that, as the number of edge nodes increases with the growing adoption of the MEC paradigm, our proposed model transfer strategy will help reduce training time and energy consumption. TransMUSE will only need to revisit cluster membership and will circumvent the need to persistently collect traffic data in each district. 5 Related work Network traffic prediction is critical to network resource management, optimization and QoS improvement. While this topic has received a lot of attention over the recent years, aspects including service-level traffic forecasting and predicting with low computational overhead have been largely overlooked. Here, we summarize the most relevant work related to our contribution. 5.1 Traffic forecasting The main approaches to time sequence prediction are State Space Models (SSMs) and sequential models that frequently use deep learning (DL) [29]. The most representative SSMs are Auto-Regressive Integrated Moving Average (ARIMA) models and variants of these, which have been widely adopted for mobile traffic forecasting [30], [31], [32]. Their major drawback is that they require manual parameter selection on a sequence-by-sequence basis. In addition, they perform poorly when inputs exhibit high variability. DL has made advances in multiple domains, with Long–Short Term Memory (LSTM) models proven to be superior to traditional models such as ARIMA when predicting wired and wireless traffic [18], [33], [34], [35]. Given that spatial correlations exist between traffic generated at different base stations in wireless networks, LSTM models have been combined with Convolutional Neural Networks (CNNs) to tackle this problem. Zhang et al. proposed a ConvLSTM model to predict multi-service mobile traffic [7] and a graph-sequence spatio-temporal model is introduced in [36] to forecast cellular traffic demand. More recently, attention and transformer architectures demonstrated the ability to handle long sentences in the NLP domain, which subsequently led to their adoption in time series forecasting tasks [6], [29]. However, none of these prior works builds on the observation that spatial correlations are weak in wired networks and correlations among services matter most. 5.2 Edge model transfer As edge computing is getting traction, there have been several research projects focusing on cloud–edge model training based on collaborative learning. He et al. design a collaborative global–local learning scheme that leverages the generalization capability of the global model and the personalization ability of local models to boost the training performance of a graph attention spatio-temporal network (GASTN) for city-wide mobile traffic prediction [6]. Yan et al. propose COLLA, a collaborative learning framework that allows devices and the cloud to learn collectively user locations [37]. Zhang et al. design a collaborative cloud–edge computation method for driving behavior modeling, which trains and prunes common models in the cloud and conducts transfer learning at the edge [38]. Cartel is proposed in [39] for cloud–edge collaborative learning, aiming at distributing and updating machine learning models across geographically distributed edge clouds. These works are mostly set on the premise that there exists plenty of data in the cloud to train global models. Edge–edge collaboration, in scenarios where data is largely available only at the network edge, has received less attention. Further, the cost of data transfer overhead has been thus far overlooked, which is non-negligible for network operators. 6 Conclusion In this paper, we tackled network traffic prediction in multi-service edge networks with spatially heterogeneous demands. We proposed TransMUSE, a framework that groups edge nodes into cohorts and trains transformer-based (TMTPN) models at reference locations, which can be transferred within cohorts without any adaptation. By means of extensive experiments with real-world data, we demonstrated TransMUSE’s forecasting performance is comparable with that of training individual models with local data at each node. We further propose WK-means, a service clustering routine, which allows to reduce the number of TMTPN models to be maintained for forecasting, based on service similarities. All of these facilitate accurate short- and long-term multi-service traffic prediction with reduced measurement and training costs, which is essential for fine-grained network management. CRediT authorship contribution statement Luyang Xu: Conceptualization, Methodology, Formal analysis, Experiments, Writing – original draft, Review and editing. Haoyu Liu: Methodology, Writing – review. Junping Song: Data cooperation and analysis, Writing – review. Rui Li: Conceptualization, Writing – review. Yahui Hu: Conceptualization. Xu Zhou: Conceptualization, Resources, Supervision, Project administration. Paul Patras: Conceptualization, Writing – review and edit, Supervision, Project administration. Luyang Xu is a Ph.D. candidate in Computer Network information Center, Chinese Academy of Sciences. He receives his B.Sc. degree in computer science from Northeast Forestry University in 2016. His research interests focus on the network edge traffic analysis, including traffic classification, traffic prediction and unknown traffic identification, which introduces the artificial intelligence techniques to solve the problems existing in network domains. Haoyu Liu received the B.Sc. degree in computer science from South China University of Technology and the University of Edinburgh in 2019. He is currently pursuing the Ph.D. degree in the School of Informatics at the University of Edinburgh. His research direction is leveraging machine learning and deep learning algorithms to solve network security and privacy issues, including intrusion detection and anti-censorship. Junping Song received Ph.D. degree from the institute of acoustics of the CAS in 2013. Then she joined the Institute of Software CAS as an assistant research fellow. In 2016, she joined GTCOM as a researcher. She currently works for the Computer Network Information Center, CAS. Her research interests are network artificial intelligence, network architecture, etc. Rui Li is a machine learning researcher at Samsung AI Center in Cambridge, UK. She is interested in communications theory, networking and machine learning. She obtained her Ph.D. from School of Informatics, the University of Edinburgh, advised by Dr Paul Patras. She was the recipient of the Brendan Murphy Memorial Prize in 2018. Rui received her M.Sc with Distinction from University of Leicester, where she was awarded the Best Student Prize from Department of Engineering. Dr. Yahui Hu is currently an associate professor in China University of Mining and Technology (Beijing). She received her doctor's degree from Beijing university of Posts and Telecommunications in 2009. After that, she continued researches on wireless communication network in High Performance Network Lab. During this period, she has been in charge of or participated in many national projects, such as National Natural Science Foundation, 863 Project, National Major Scientific Project, etc. She has published more than 20 papers and applied national 6 patents in the above field. Now, her research interests include quality-driven cross-layer optimized multimedia over wireless, cognitive network management and wireless network management and optimization. She has served as the technical program committee member or paper reviewer for Europe Transactions on Telecommunications, IEEE Transactions on communications, IEEE Proc. of Globecom, etc. Xu Zhou is currently a professor at the Computer Network Information Center, Chinese Academy of Sciences (CAS). He received B.S. degree and M.E degree from the Sichuan University in 2001 and a Ph.D. degree from the University of Electronic Science and Technology of China in 2005. His main study interests include future networks such as 5G and B5G. Paul Patra is an Associate Professor in the School of Informatics at the University of Edinburgh, where he leads the Mobile Intelligence Lab -- a multi-disciplinary team that pursues research at the intersection of network engineering and artificial intelligence, to improve the analysis, resilience, and management of next generation mobile systems. He is also a co-founder and CEO of Net AI, a pioneering university spinout specializing in AI-driven network analytics. He has served on the organizing committee on several conferences and workshops in his field, and advised the ITU-T Focus Group on Machine Learning for Future Networks including 5G. Paul holds M.Sc. and Ph.D. degrees from Universidad Carlos III de Madrid (UC3M) and he was the recipient of a prestigious Chancellor's Fellowship awarded by the University of Edinburgh. Data availability The authors do not have permission to share data. Acknowledgments This work was partially supported by the 10.13039/501100001809 National Natural Science Foundation of China (Grant No. U1909204), the 10.13039/501100012166 National Key R&D Program of China (Grant No. 2020YFB1-806002), 10.13039/501100004826 Beijing Natural Science Foundation , China (No. 4202082), the 10.13039/501100004543 China Scholarship Council and 10.13039/501100004739 Youth Innovation Promotion Association of Chinese Academy of Sciences (2021168). ==== Refs References 1 Fang Y. Diallo A. Zhang C. Patras P. Spider: Deep learning-driven sparse mobile traffic measurement collection and reconstruction 2021 IEEE Global Communications Conference 2021 Institute of Electrical and Electronics Engineers (IEEE) 2 Chang Y.-Y. Sun F.-Y. Wu Y.-H. Lin S.-D. A memory-network based solution for multivariate time-series forecasting 2018 arXiv preprint arXiv:1809.02105 3 Huang C.-W. Chiang C.-T. Li Q. A study of deep learning networks on mobile traffic forecasting IEEE PIMRC 2017 4 Zhang C. Patras P. Long-term mobile traffic forecasting using deep spatio-temporal neural networks ACM MobiHoc 2018 231 240 5 Wang X. Zhou Z. Xiao F. Xing K. Yang Z. Liu Y. Peng C. Spatio-temporal analysis and prediction of cellular traffic in metropolis IEEE Trans. Mob. Comput. 18 9 2018 2190 2202 6 He K. Chen X. Wu Q. Yu S. Zhou Z. Graph attention spatial-temporal network with collaborative global-local learning for citywide mobile traffic prediction IEEE Trans. Mob. Comput. 2020 7 C. Zhang, M. Fiore, P. Patras, Multi-service mobile traffic forecasting via convolutional long short-term memories, in: IEEE International Symposium on Measurements & Networking (M&N), 2019. 8 Vaswani A. Shazeer N. Parmar N. Uszkoreit J. Jones L. Gomez A.N. Kaiser Ł. Polosukhin I. Attention is all you need Advances in Neural Information Processing Systems 2017 5998 6008 9 R. Singh, M. Fiore, M. Marina, A. Tarable, A. Nordio, Urban Vibes and Rural Charms: Analysis of Geographic Diversity in Mobile Service Usage at National Scale, in: The World Wide Web Conference, 2019, pp. 1724–1734. 10 J. Shen, Y. Qu, W. Zhang, Y. Yu, Wasserstein distance guided representation learning for domain adaptation, in: Thirty-Second AAAI Conference on Artificial Intelligence, 2018. 11 H. Liu, X. Gu, D. Samaras, Wasserstein GAN with quadratic transport cost, in: Proceedings of the IEEE/CVF International Conference on Computer Vision, 2019, pp. 4832–4841. 12 Arjovsky M. Chintala S. Bottou L. Wasserstein generative adversarial networks International Conference on Machine Learning 2017 PMLR 214 223 13 Wasserman L. Optimal transport and wasserstein distance 2020 http://www.stat.cmu.edu/~larry/=sml/Opt.pdf [EB/OL] 14 Dong X.-L. Gu C.-K. Wang Z.-O. Research on shape-based time series similarity measure International Conference on Machine Learning and Cybernetics 2006 IEEE 1253 1258 15 Devlin J. Chang M.-W. Lee K. Toutanova K. Bert: Pre-training of deep bidirectional transformers for language understanding 2018 arXiv preprint arXiv:1810.04805 16 X. Pan, Z. Xia, S. Song, L.E. Li, G. Huang, 3D object detection with pointformer, in: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2021, pp. 7463–7472. 17 Xu M. Dai W. Liu C. Gao X. Lin W. Qi G.-J. Xiong H. Spatial-temporal transformer networks for traffic flow forecasting 2020 arXiv preprint arXiv:2001.02908 18 Trinh H.D. Giupponi L. Dini P. Mobile traffic prediction from raw data using LSTM networks International Symposium on Personal, Indoor and Mobile Radio Communications PIMRC 2018 IEEE 1827 1832 19 Vinayakumar R. Soman K. Poornachandran P. Applying deep learning approaches for network traffic prediction 2017 International Conference on Advances in Computing, Communications and Informatics ICACCI 2017 IEEE 2353 2358 20 Cerliani M. Time Series Forecasting with Graph Convolutional Neural Network 2020 https://towardsdatascience.com/time-series-forecasting-with-graph-convolutional-neural-network-7ffb3b70afcf [EB/OL] 21 Kipf T.N. Welling M. Semi-supervised classification with graph convolutional networks 2016 arXiv preprint arXiv:1609.02907 22 Grattarola D. Spektral for Graph Deep Learning 2021 https://github.com/danielegrattarola/spektral [EB/OL] 23 Bahdanau D. Cho K. Bengio Y. Neural machine translation by jointly learning to align and translate 2014 arXiv preprint arXiv:1409.0473 24 Wu Q. He K. Chen X. Yu S. Zhang J. Deep transfer learning across cities for mobile traffic prediction IEEE/ACM Trans. Netw. 2021 25 Krell M.M. Wehbe B. A first step towards distribution invariant regression metrics 2020 arXiv preprint arXiv:2009.05176 26 Ren X. Gu H. Wei W. Tree-RNN: Tree structural recurrent neural network for network traffic classification Expert Syst. Appl. 167 2021 114363 27 Tslearn for the analysis of time series 2017 https://tslearn.readthedocs.io/en/stable/ [EB/OL] 28 Shahapure K.R. Nicholas C. Cluster quality analysis using silhouette score 7th International Conference on Data Science and Advanced Analytics DSAA 2020 IEEE 747 748 29 Wu N. Green B. Ben X. O’Banion S. Deep transformer models for time series forecasting: The influenza prevalence case 2020 arXiv preprint arXiv:2001.08317 30 Adas A. Traffic models in broadband networks IEEE Commun. Mag. 35 7 1997 82 89 31 Shu Y. Yu M. Yang O. Liu J. Feng H. Wireless traffic modeling and prediction using seasonal ARIMA models IEICE Trans. Commun. 88 10 2005 3992 3999 32 Sultan K. Ali H. Zhang Z. Call detail records driven anomaly detection and traffic prediction in mobile cellular networks IEEE Access 6 2018 41728 41737 33 Rago A. Piro G. Boggia G. Dini P. Multi-task learning at the mobile edge: An effective way to combine traffic classification and prediction IEEE Trans. Veh. Technol. 69 9 2020 10362 10374 34 Abbas N. Zhang Y. Taherkordi A. Skeie T. Mobile edge computing: A survey IEEE Internet Things J. 5 1 2017 450 465 35 Hua Y. Zhao Z. Li R. Chen X. Liu Z. Zhang H. Deep learning with long short-term memory for time series prediction IEEE Commun. Mag. 57 6 2019 114 119 36 Fang L. Cheng X. Wang H. Yang L. Mobile demand forecasting via deep graph-sequence spatio-temporal modeling in cellular networks IEEE Internet Things J. 5 4 2018 3091 3101 37 Y. Lu, Y. Shu, X. Tan, Y. Liu, M. Zhou, Q. Chen, D. Pei, Collaborative learning between cloud and end devices: an empirical study on location prediction, in: Proceedings of the 4th ACM/IEEE Symposium on Edge Computing, 2019, pp. 139–151. 38 X. Zhang, M. Qiao, L. Liu, Y. Xu, W. Shi, Collaborative cloud-edge computation for personalized driving behavior modeling, in: Proceedings of the 4th ACM/IEEE Symposium on Edge Computing, 2019, pp. 209–221. 39 H. Daga, P.K. Nicholson, A. Gavrilovska, D. Lugones, Cartel: A system for collaborative transfer learning at the edge, in: Proceedings of the ACM Symposium on Cloud Computing, 2019, pp. 25–37.

0

PMC9753913

NO-CC CODE

2022-12-16 23:26:22

no

Comput Netw. 2023 Feb 11; 221:109518

utf-8

Comput Netw

10.1016/j.comnet.2022.109518

oa_other

==== Front Technol Forecast Soc Change Technol Forecast Soc Change Technological Forecasting and Social Change 0040-1625 0040-1625 Elsevier Inc. S0040-1625(20)31357-3 10.1016/j.techfore.2020.120531 120531 Article Lockdown and sustainability: An effective model of information and communication technology Shareef Mahmud A. a Dwivedi Yogesh K. b⁎ Wright Angela c Kumar Vinod d Sharma Sujeet K. e Rana Nripendra P f a School of Business & Economics, North South University, Dhaka, Bangladesh b Emerging Markets Research Centre (EMaRC), School of Management, Swansea University Bay Campus, Fabian Way, Swansea, SA1 8EN, United Kingdom c Department of OPD, School of Business, Cork Institute of Technology, Ireland d Sprott School of Business, Carleton University, Ottawa, Canada e Information Systems & Analytics Area, Indian Institute of Management Tiruchirappalli, India f School of Management, University of Bradford, Richmond Road, Bradford, BD7 1DP, United Kingdom ⁎ Corresponding author. 5 1 2021 4 2021 5 1 2021 165 120531120531 5 6 2020 27 11 2020 11 12 2020 © 2020 Elsevier Inc. All rights reserved. 2020 Elsevier Inc. Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. Covid-19, a corona virus, has maintained its momentum in spreading among communities. In this context of social crisis, this study seeks to identify the reasons for the partial failure to fulfill the intended goal of lockdown, and to formulate an inclusive behavioral model reflecting comprehensive human behavior and social psychology. In order to answer the research questions, this study has conducted extensive interviews among individuals who were targets of the lockdown system. From this exploratory and qualitative investigation, researchers have recognized four paradigms as the key to understanding human behavior and social psychology in violating lockdown as a social isolation system during this period of crisis. The identified parameters depicting social behavior are: Derogation and Argument (SDA), Tangible Need and Deficiency (TND), Intangible Desire and Expectancy (IDE), and Evaluation of Benefit and Loss (UBL). Finally, as a comprehensive guideline, a grounded theory of the social behavior ‘paradigm for lockdown violation (PLV)’ is explored as the reason for the violation of the social system. Keywords Lockdown Social isolation Social crisis Sustainability Social psychology Human behavior ==== Body pmc1 Introduction The year 2020, from its inception, has made the word ‘Lockdown’ a familiar term and universal concept. From the start of January, the pandemic Covid-19 has had a devastating effect worldwide. It has accelerated the use of the word ‘lockdown’ so profoundly that it warrants research as to whether any word in recent centuries has been used more often than this life-altering word is presently articulated in common, scientific and governmental parlance. Research studies in all areas typically identify whether the sample effect is categorized as respondents from developed countries and/or developing countries (Weinstein, 1994). Although, for many studies, standardization strongly recommends ignoring country differences, generally, for research either qualitative or quantitative, country differences, broadly divided as developed and developing countries, are considered substantially (Porter, 1998; Shareef et al., 2013). In marketing, market segmentation is also a common parameter which also necessitates consideration of demographical characteristics and cultural differences based on education, race, class etc., (Kumar et al., 2011). However, research on Covid-19 is a classic example where any market segmentation can be realistically ignored based on either the global economy (developed and developing countries) or demographic factors inside a country. People and/or information are restricted under lockdown to enter into or exit from a particular area or building or room or from any computing device (Schept, 2013; Wang et al., 2020). So, lockdown is a preemptive action where all people (or information) are confined in a certain location, area, compound, building, and/or room partially or completely under a regulation assigned by legal authority for a particular time either definite or indefinite. Therefore, under the protocol of lockdown, a specific room, building, compound, community, area, city, state, and/or country is kept closed, and nobody inside can exit and nobody from outside can enter without proper permission. This refrainment can be partially or completely enforced (Lau et al., 2020). During this period, all people are prohibited from leaving or entering other than the law enforcing authority. Lockdown can be one of several types: emergency lockdown for physical security; and preventive lockdown for protection from possible physical or mental damage. Lockdown can be partial where people are permitted to come out from the confinement for a while and an outsider can enter with permission (Wang and Wu, 2018; Zhong et al., 2020). It can be complete lockdown where nobody inside can exit and nobody from outside can enter in that confined place other than special permission from the law enforcing authority for emergency or special needs. Lockdown can be fully restricted under which entrance and exit is completely prohibited under any circumstances. Under the present scenario of Covid-19, lockdown, either partial, complete, or fully restricted, can be defined as a preemptive action plan designed with certain protocols to restrict transmission of the virus from outside or from inside by closing a certain building and/or a social community (designated by location, area, city, state, country) and all human beings and their pets are to stay within that boundary (Schept, 2013; Wang et al., 2020). They are not permitted to come out from that restricted place to other communities and people from other communities are also prohibited from entering that specific location without the permission of the law enforcing authority. It is both an emergency and preventive lockdown implemented by the government at any level. The present scenario, realistic outcomes, and its implied significance needs to be examined from the distinct perspective of various countries. Under this context, the consequences of lockdown can be addressed, analyzed, and comprehended in several settings. For instance, in the USA, different states have imposed local lockdowns regulated by the relevant state authorities. China, the originator of this pandemic (initially it was considered as an epidemic) which was suffering almost alone in the first month of 2020, placed a lockdown in Wuhan only, the epicenter of the coronavirus outbreak, in a bid to prevent any further spread of the disease where the population is almost 11 million (The Guardian, 2020; Zhong et al., 2020). Japan was reluctant to put any countrywide lockdown in place initially in order to maintain the country's economy. In Italy, due to the severe expansion of the Covid-19 infection, a nationwide lockdown was put in place to address the devastating conditions, although restrictions were frequently not observed by many citizens. The Italian prime minister has eased some of its stringent nationwide lockdown restrictions in May (Ellyatt, 2020). France has put a nationwide lockdown in place and their president is expecting to relax lockdown while maintaining partial restrictions (Ellyatt, 2020). Similarly, a complete nationwide lockdown was imposed in Spain and the Republic of Ireland. These measures became the new normal for citizens worldwide, in both developed and developing countries. Now the obvious question and of paramount concern is the effectiveness of the regulatory lockdown system imposed by almost all the governments in the world. From the conceptual definition, lockdown is intended to keep people of a particular location isolated from other communities (Figueiredo et al., 2020). So, it is focused on separating people. Since lockdown was often imposed, monitored, and regulated by the government authorities in extremely strict fashion without allowing citizens any scope to exercise their citizenship rights, in several countries like China, North Korea and, to some extent South Korea, lockdown worked effectively (The Guardian, 2020). However, due to the scope of so-called democratic rights for free movement, flexible government policy, the relaxed performance of law enforcing authorities, and/or governments’ hidden agenda to allow economic activities to maintain economic growth, in several countries all over the world, lockdown, whether partial, complete or fully restricted, did not work to meet expectations and protect community transmission of the virus (Lau et al., 2020; Wang and Wu, 2018; Zhong et al., 2020). Fundamentally, in some countries, lockdown has failed to achieve its desired goal due to the unwillingness of many people to strictly follow government regulations regarding lockdown. Now, if the public do not want or are not able to abide by rules of separation, and frequently come out from their closed loop and interact with people in other locations, the intended purpose of lockdown cannot be effective. Community spreading will not be controlled and ultimately lockdown will fail. According to the world news media, for instance the BBC, CNN, CBC, Al Jazeera, Reuter, AFP etc., it is obvious that the effectiveness of lockdown gradually reduced as many people from various communities did not perceive it as necessary to be locked down for days (Lau et al., 2020; Wang et al., 2020). They simply ignored or underestimated the implied significance of separation imposed by lockdown and broke the physical confinement of home (Wang et al., 2020; Wang and Wu, 2018). For the sustainability of any society, i.e., for the simplest form of existence, social members with standard social norms depicting a notion of civilization should be alive. The present lockdown, worldwide, is extending and expanding only to ensure the simplest existence of civilization; nevertheless, it must confirm and should not preclude the human urge of social capital, an inherent and eternal urge of community development (Bunker, 2020; Luciano, 2020; Venkatesh, 2020). Heuristically, on one side of the coin, it is needed to place and ensure lockdown for sustainability (Kumar and Managi, 2020; Nakamura and Managi, 2020; Yoo and Managi, 2020). On the other side of the coin, illustrating human urge, appeal, and prevalent nature, the autonomous trend of willful mixing in the form of community development cannot be denied (Dwivedi et al., 2020; Katafuchi et al., 2020; Kurita and Managi, 2020). Ignoring the second side of the coin, lockdown cannot be successful or bring the desired goal which all the governments are hoping for their citizens. However, presently, in all spheres of the world, lockdown has limited success. Revealing the underlying reasons of apparent failure of lockdown to ensure the desired target is extremely important for the sustainability of civilization, and, above all, human life. Therefore, in this present scenario, this research has a twofold objective:• Addressing the fundamental purpose of lockdown and identifying the reasons for the partial failure to fulfill the desired target of lockdown. • Now based on the answer of the first objective, this research has set its second objective to be the formulation of an inclusive behavioral model reflecting comprehensive human psychology in the context of any social crisis. It is optimistically expected that this research will contribute extensively to the literature of psychology and public administration. It can also help to predict human behavior during any period of crisis similar to the pandemic of Covid-19. It should also have an enormous impact on designing a sustainable society approach to abiding by government rules and regulations during any period of emergency. The next section of this study reviews the literature on social psychology and human behavior and sustainability. Then a theoretical framework is developed. The following section proposes the research design. Subsequently, the results and discussions of this study will be presented. Then the theoretical and managerial implications of this study are explained in the following sections along with the limitations of this study and future research directions. Finally, a conclusion is outlined. 2 Literature review: introduction According to the Director General of the World Health Organization (WHO) outlining the lockdown protocol, “these measures are the best way to suppress and stop transmission, so that when restrictions are lifted, the coronavirus doesn't resurge,"(WHO, 2020). The Director General of WHO further warned countries that, “lifting restrictions too quickly could lead to a deadly resurgence” (WHO, 2020). Regarded as a vital area of research, this study addresses the WHO's guidelines that a deadly spreading and community infection can be only managed and controlled if people maintain the strong regulation of lockdown. Examining recent results either from developed or from developing countries, people are breaking the notion of lockdown by leaving their confined homes, mixing with community, and maintaining socialization. Why is this happening? This is a critical issue for the existence of society, for countries and the whole world. It is logical to underpin and analyze this issue from two perspectives: externally, from the conceptual paradigm of sustainability and, internally, from the psychological theory of human beings. 2.1 Sustainability First, the theoretical dogma of sustainability is examined. Several researchers have explained that any system will sustain itself if it can fulfill people's certain intrinsic and extrinsic motivation (Lozano, 2008; Robinson and Tinker, 1998; Rosner, 1995) for that system. Researchers from multidisciplinary fields (Doppelt, 2003; Lozano, 2008) have agreed that people perceive any system and its development favorably if it has the ability to fulfill three basic requirements, such as social, economic and environmental aspects (Senge, 1999). Theorizing the socioeconomic paradigm with the contextual environment dynamic and the inter-related relationship of economy, social life and environment is the powerful issue for sustainable development (Archibugi et al., 2013; Dresner, 2002; Lozano, 2008; Montabon et al., 2016; Rey-Martí et al., 2016; Ribeiro-Soriano and Salvador, 2009; Roseland, 2000). That means, if any system can meet people's requirements of environmental, economic and social deficiencies, it is more likely to be sustainable (Dresner, 2002; Rees, 2002; Robinson and Tinker, 1998). Therefore, it can be postulated, with the prerequisite conditions, that, for the sustainability of any society, it should maintain the standard of social norms and human equality with the standardization of civilization, and it is rooted on the pillar of three fundamental aspects: social dimension, environment dimension, and economic dimension (Cairns, 2004; Kirkby et al., 1995; Lozano, 2008; Montabon et al., 2016; Roseland, 2000). Pragmatically, highlighting the social appeal for existence keeping equity, it is needed to establish and warrant a social system which can survive, exist, and move forward with prevailing norms and standards (Diesendorf, 2000; Langer and Schoon, 2003). 2.2 Human behavior theory At this stage, it is necessary to address the psychological behavior of human beings. For predicting human behavior, one famous and popular psychological domain is the theory of planned behavior (Ajzen, 1991). Focusing on the root of this theory, human behavior is examined through the deterministic influence of their attitudes (self-developed perception/ feelings), subjective norms (approval of referent group), and perceived behavior control (external catalyst to facilitate or hinder performance of the behavior). Shedding light on the conceptual significance of this behavioral theory, people will abide by the regulation of lockdown (to be confined in a restricted area and separated from community) if they have certain internal beliefs which may positively pursue their attitude to execute the central theme of lockdown. Also, they must be influenced by their associates (reference group) to maintain confinement in their residential location and isolation from all personal, social, economic, and community-based activities (Pachidis et al., 2019). Following the idea of lockdown also depends on several external factors (perceived behavioral control), for instance, the possibility of punishment if it is not strictly adhered to, the scope of getting rewarded if it is maintained, and the tangible possibility of sufferings by transmission of the virus if the lockdown is broken. On the contrary, perceived behavioral control may have a negative impact (wishing to break the system from isolation) if it is substantially pressurized by some negative factors, such as losing a job, missing economic opportunities, and attraction for social interactions (although, in this scenario it is a negative factor) (Kaburlasos and Vrochidou, 2019) etc. Analyzing the balance theory (Heider, 1958), it is well recognized that human behavior is greatly influenced by celebrity endorsement. It means, if someone likes a celebrity, that person is potentially disposed to wish to follow that celebrity's behavior. This liking can be extended by developing social networks which finally can develop a group behavior. The dangerous side of human psychology which is unveiled by this theory is that human minds might oppose any systems and enthusiastically strive to disobey any regulations if they perceive the administrator (celebrity) negatively and unfavorably. That means, reflecting the central theme of this psychological theory, the human mind is persuasive to any regulation either favorably or unfavorably depending on the evaluation of the source derogation (Alvaro and Crano, 1996; Wood, 2000). Before completing a brief, summative, and integrative view of human psychology, focused on the uneven path of human decisions, an economics concept named prospect theory (Kahneman and Tversky, 1979) should be analyzed and contrasted. This theory was developed by Daniel Kahneman and Amos Tversky in 1979 to challenge and necessitate the modification of expected utility theory (Neumann and Morgenstern, 1953). This prospect theory is focusing, contrasting, and integrating the conjoint effect of loss aversion and an asymmetric form of risk aversion (Tversky and Kahneman, 1992). Clearly the human mind uses its mental ability to diagnose a situation and relatively evaluate the possibility of potential losses and potential gains from the occurrence of that event. Consequently, and heuristically, people tend to choose their final behavior based on the relative betterment from a specific situation by weighing potential gains and potential losses incurred. 3 Theoretical framework Given the novel aspect of this study, this unprecedented new area lacks theoretical support to develop a definite theoretical underpinning to advance this study. However, from the previous analysis and illustration, some background paradigms of theoretical structure can be assumed:Ø Since a community-based mass of people is closely associated with the process of lockdown, this isolation and refrainment from all kinds of community-based interaction should be examined considering overall human behavior — their cognitive and affective attitudes. Ø During this lockdown period, people are closely confined in their specific residential locations. Therefore, they are continuously missing their social urge and emotions to share with others. Consequently, this issue has a strong social aspect. Ø Rooted in financial deficiency, the economic dimension is an important factor when examining lockdown. Behavioral economists (Camerer, 1989; Cohen et al., 1987; Tversky and Kahneman, 1992) have argued for many years that human behavior is closely intertwined with their financial requirements. Theorists working on organizational motivation (Donaldson, 1990; Dulaimi et al., 2003) also agreed on the fact that economic deficiency can mold and aggravate human behavior in certain conditions. Ø Human behavior in occupations, professional activities, urbanization and industrialization — i.e., nationwide all economic activities have a strong impact on environmental degradation. In recent years, environmental scientists have strong recommendations to save the climate by taking several measures which can ensure reduction in carbon emissions. Ø Human psychology is a complex issue and it has multidimensional aspects, some of which have driving and pursuing effects, and several of which have inhibiting and controlling thrust (Bardecki, 1984; Bowonder and Linstone, 1987; Cohen et al., 1987; Dwivedi et al., 2016; Shareef et al., 2020b). Combining, comprehending, and integrating all these effects, compound behavioral patterns can be speculated on and predicted. Therefore, investigating the objectives of this exploratory study should be based and manifested on primary and generic areas of human behavior and decision-making patterns. Integrating the central concept of the aforementioned psychological, behavioral, and economic decision-making theories and the human urge for social capital in the light of sustainability, and contrasting these with the present scenario of lockdown and human behavior, several driving parameters which can be used to develop a theoretical framework for predicting the behavioral response under this emergency event can be discovered (transmission of Covid-19) as follows:1 It is an eternal human urge to gather in the community and continue socialization. 2 Pragmatically, appeal for social capital is an absolute feeling of social gains. 3 Permanency of any event should be looked at from the probable benefits of three dimensions: environmental, social, and economic. 4 The human mind explores gains from any events considering the isolated benefits derived from or for environment, society, and economy. 5 For sustainability, quality of environment, social benefits, and economic benefits must be fulfilled. 6 In any social conditions, human beings conform to the norm, if they hold a favorable attitude to the system. 7 Human behavior is not only an outcome of one's own feelings; it is also influenced by society's emotional pressure on them. 8 External factors also have a conjoined impact on controlling their final behavior 9 Images and derogations of reference members and arguments and counterarguments play a vital role in shaping the response of humans. 10 Human beings, to find comfort in their mind, always try to align their attitudes and behavior. 11 People give priority to the stress derived from the society and community. 12 People respect their desire if the situation is favorable. 13 If any situation is unbearable, people try to escape from the situation. 14 It is an inherent tendency of the human mind to evaluate the final and concurrent impact of both loss and gain relatively. 15 If the absolute benefit is greater than the probable occurrence of loss, people behave accordingly to grasp benefit. 4 Research design This study sets out its twofold objectives: firstly, to analyze and discover the reasons for violating the lockdown system worldwide. Then, based on the identified reasons, contrasting these with generic human behavior by shedding light on social psychology and sustainability, the second objective is developing a grounded theory of disobedience within any social system, reflecting comprehensive and synergistic pictures of human behavior. To grasp the fundamental essence of the two objectives, as an extraordinary exploratory investigation having weak literature support, the study has designed its methodology to conduct extensive qualitative study among the major stakeholders. As the exploratory investigation was underpinned by the semi-structured theoretical framework (rather it should be regarded as the paradigms of the outline of theoretical framework), this study is fundamentally and primarily based on conducting several queries among the principal stakeholders through qualitative study. The design of this qualitative study is divided into two phases. In the initial phase, the stakeholders of this study were identified to reach the objectives by getting appropriate answers. More specifically, in the first phase, the groups who have association and functions in the entire lockdown system as the implementers, executors, observers, and/or facilitators were recognized. Since the study and the consequent development of a theoretical base is primarily dependent on the response of the members who are closely involved in this mechanism of lockdown, identification and recognition of major stakeholders in this lockdown system is important. To reach this goal, the study has followed two distinct but connected procedures:1 Observation of the scenario in several countries derived from the broadcasting of electronic media, searching in social media for interactions and opinions of different social members, and the reading of newspapers 2 Open discussion and brainstorming with potential social members having expertise on social incidents and human behavior in focus groups. Consequently, five university professors of a leading private university in Bangladesh, two from sociology departments, two from psychology departments, and one from organizational behavior (management department) were invited to a one-hour discussion. The second phase of this study was conducted in Bangladesh where maintaining social isolation is a serious challenge. From the first phase, three broad categories of stakeholders were identified as per their role, responsibility, and function. These three categories are shown in the first column of Table 1 . Now again under each category, different types of stakeholders are mentioned in the second column of Table 1. All these stakeholders were selected as the sample. In the second phase, more detailed extensive interviews were conducted among the stakeholders identified in the first phase. From each group of stakeholders, several people were interviewed as mentioned in the third column of Table 1. The interviews were not conducted based on any specific structured questionnaire. The respondents were given complete freedom to provide their perceptions about the reasons to violate lockdown and not to abide by social isolation. The interview time varied from twenty minutes to one hour. The number of respondents from each group was selected based on their importance, involvement, priority, role, and responsibility. Considering this, the following is the number from each group and their realistic involvement and importance shown in Table 1 (Total interviews conducted =76).Table 1 Involvement of stakeholders in lockdown. Table 1Category Stakeholder Number of Respondent Role and Priority Lockdown Breaker People uninfected by the virus Covid-19 30 They are violating the lockdown most frequently. Each time they might be infected by others Personally, uninfected but family member(s)/friend(s) infected by Covid-19 10 They are very vulnerable for community transmission since they might be in very close contact with someone already infected. Presently suffering from Covid-19(mild) 2 They are extremely sensitive and risky for community transmission Infected by Covid-19, but now recovered 2 They perceive themselves risk-free from two-way transmissions; however, it is still not proven. So, their socialization can be detrimental Lockdown Implementer Government administrator (Regulatory authority) 5 Focusing on all aspects of social, behavioral, and economic need, they prepare regulations and plan for lockdown and isolation. They have statistical data about human needs and behavior Law enforcing authority (Police and Army) 5 They are the field supervisors and implementers. They are in close contact with the violators of lockdown, talk to them, discuss, and even forcefully try to control and refrain them from socialization Local elected members 5 They are responsible to guide people, discuss with them about their problems, requirements, expectations, and plan, and implement social isolation in their respective community Lockdown motivator Physicians and other health service workers 5 They can understand and collect patients’ view and way of infection and transmission. Relief distribution authorities (government and private— NGO) 10 They realistically realize the lockdown violators’ view, need, and urgency. They know and understand the general public's expectation and deficiencies Professional psychologists/ sociologists 2 They have theoretical knowledge to analyze, understand, and recommend for the public's intentions, motives, attitudes, and behavior Shedding light on the nature, role, responsibility, and authority of the respondents who were interviewed, there are broadly three categories of primary stakeholders who are active and/or passive members of the lockdown system. First category, representing the general public, is the possible breaker and violators of the lockdown system (lockdown breaker) for any reasons — logical or unjustified. Second category is the official authority to prepare regulations, enforce, endorse, and execute those regulations. They also strictly control the first group as the lockdown implementer. Third category is the lockdown motivator. They are responsible to provide direct treatment and/or other supplementary services for livelihood and mental stability. So, they have scope to motivate human behavior in favor of lockdown. Since this is an entirely exploratory study where concrete theoretical framework is absent, this present investigation followed a systematic procedure to ensure authentication of the study regarding this qualitative investigation through interviews, filed observation, and literature review. For this method of study, justification of data source, collection method, and investigators’ reliability and validity have potential value and, thus, this study conducted the entire investigation maintaining the basic principles of triangulation methods (Moon, 2019; Patton, 2002; Shareef et al., 2020a). This qualitative study enabled data source triangulation by collecting information from several categories of stakeholders who have influence and interference in the social system of lockdown. Information was collected following different distinct and appropriate methods to validate method triangulation. Depending on the vulnerability, availability, and schedule, data was collected following several methods: a) Face-to-face interviews; b) Telephone interviews; c) Video interviews; d) Discussion; e) Filed observation. To avoid investigator bias and preconceived inclination to any outcome, information was collected by four persons (researchers and research assistants). In this way, personal bias was carefully avoided, and investigator triangulation was confirmed. This study is very much unstructured, since it does not have enough theoretical reference. The collected data and collection procedure, at this present moment, is very risky and sensitive. As a result, to ensure validity and reliability of the outcome, information, i.e., the verbal statement collected from this qualitative method was rearranged, reorganized, restructured, and converted into common discourses and conceptual paradigms keeping the consistency in respect to generic meaning and significance and implied implications. 5 Results and discussions Given the variations inherent in raw statement and verbal expression, information extracted from the three categories has different focus, motive, reasoning, and style. However, while converting these into small pieces to unfold and synthesize potential significance, commonalities by contrasting potential analogy, systematic direction, and strong theoretical guidelines have been observed among the precise statements of the three categories. Basic steps of discourse analysis were followed to come up with the precise identifications of reasons for violating social isolation and government regulations during lockdown. Under the three broad groups who are somehow related with lockdown system (see Table 1), all the stakeholders were informed about the context of the interviews. Following the accumulation process of statements under each category, information from each category was organized. In this aspect, as the reference of conversion of long statements into small pieces to reveal justified and structured paradigms of human behavior systematically, the principles of matrix thinking (Patton, 2002) were applied with appropriate coding of material. Above all, the literature review compiled so far (Daly and Wilson, 1999; Rosner, 1995; Shettleworth, 1998; Wood, 2000; Zhonge et al., 2020) was contrasted to further justify the outcome of human behavior. Cultural references were also verified for each group of stakeholder. Finally data was interpreted. Organizational sense-making through cognitive schemas (based on past knowledge) and situated cognition (interpreted considering contextual situation), behavioral requirements considering psychological status derived and recommended by psychological theories (Wood, 2000), and paradigms of sustainability (Cairns, 2004) were also applied to contrast the findings with reliable and generalized paradigms of cause and effect. The findings which indicate the possible reasons of violation and failure of lockdown are shown in Table 2 . Findings listed in Table 2 can fulfill the first objective of this exploratory research.Table 2 Reasons of violation and failure of lockdown. Table 2Category Reasons based on Self-Judgment and Explanations Lockdown Breaker • Boring to stay continuously at home • For at least a certain time, need for social gathering • Forceful confinement at any place is not tolerable • Stay with family members continuously 24/7 is monotonous • Endless viewing of TV, movies etc., is not enjoyable • Prefer to interact with community, friends, and colleagues • Prefer to enjoy nature and breathe in the open air • Staying at home continuously can stop or obstruct income and professional engagement • Searching for a new opportunity of income • Need daily supporting facilities, such as banking, shopping grocery, medicine • Miss the regular life pattern/ habits (once it was boring, but due to embargo, it is now a dream to get that life back) • To compensate future monetary risk, search for additional income Lockdown Implementer • Habit to violate law and order and enjoy this style without any tangible benefits • Cannot realize the transmission of this disease and its devastating effect • Absence of social responsibility • People infected by someone else from the community do not care about further transmission from themselves • Create artificial excuse for socialization • Need for engagement in economic activities • Search for new job scopes to compensate deficiency in income • Family bonding is not good and enjoyable from the past experience • Release mental stress Lockdown motivator • Risk and mental agony aversion • Perceiving anxiety, earn more money for additional transactional cost and security • Behavior and attitude toward life • Image of government among public is most likely not favorable • Do not perceive the outcome and impact of disease seriously and profoundly • Get relief after own infection • Search for livelihood supports • To dispose mental stress by socialization • Interaction with nature by keeping detachment from relentless discussion about the disease • Common human urge to create some scope of variation Postulating a theoretical framework as the fulfillment of the second objective of this research, at this stage, depends on direct evaluation of the relationship between those identified general reasons with the fifteen (15) root and generic causes stated at the end of the theoretical framework section. Now after proper analysis, comparison, and comprehension of the summative findings shown in Table 2, an implied and generic analogy was revealed. From that analogy, it is quite evident that a systematic guideline for ontological paradigms can be drawn. The identified causes revealed from the three categories of primary stakeholders have quite distinct perspectives, unique speculations, and different motives in their perception of the problems. From the analysis, it is seen that, while the lockdown breaker group consider some reasons as the ‘effect’ of the lockdown, the lockdown implementer group perceive the similar reasons as the ‘cause’ of violation. It is quite interesting. However, the third group, the lockdown motivators, view the issue from a totally separate position. They have no direct role and responsibility in terms of official accountability and credibility to execute and implement the lockdown or to abide by the regulations. Heuristically, their involvement, in this interview process is like a moderator. So, their analyses have suggested both causes and effects from neutral speculation. While performing this procedure, the researchers consulted with the previously formed focus group as well. Finally, the following paradigms were identified as the generic root causes of human behavior for any response to the lockdown implemented by the government for the sake of the general public. This study has revealed and recognized these four paradigms as the fundamental contributions to understanding human behavior and social psychology in violating lockdown, a social isolation system during a world crisis. The identified parameters depicting social behavior are: Derogation and Argument (SDA), Tangible Need and Deficiency (TND), Intangible Desire and Expectancy (IDE), and Evaluation of Benefit and Loss (UBL). These parameters were defined with their conceptual perspectives in order to get and present a synergistic view of the public response and their behavior during a world health crisis. Finally, as a comprehensive guideline, a grounded theory of social behavior as the reasons for violating the social system during a crisis period, identified as the paradigm for lockdown violation (PLV) is presented. It is also argued that the sustainability of a social system does not depend solely on economic, environmental, and social dimensions. Sustainability of any social system also depends on the individual trait which is defined as behavioral motive (BM) (this is supported by Diesendorf, 2000 and Dwivedi et al., 2016). It is a distinct parameter for sustainability of a social system and is not connected with economic, social, and environmental aspects. 5.1 Source derogation and argument (SDA) The public evaluates the credibility of a source of information, along with the logic and counter logic of the discourses and regulations, and in their understanding of this evaluation, the public makes a decision to attribute trust to the source of lockdown information, initiator, announcer, regulating authority, and law enforcing authority. This theory can be structured with the conceptual definition: As the reason for a response to the framework of lockdown of social isolation, it is the generic perception of the public through the cognitive and affective evaluation of government and its associated agencies who are pursuing, imposing, and executing in respect to their credibility, qualification, and acceptability. Shedding light on this paradigm, to accept, follow, and maintain the framework of the lockdown of social isolation, the public are attempting to carefully understand, analyze, and comprehend government and its associated agencies’ (executive and regulatory authority and law enforcing authority) authenticity, credibility, trustworthiness, motives, and justification of their arguments to develop and pursue their attitude at the outset. While doing this, they do not find any inclination to accept, follow, and maintain the framework of lockdown of social isolation. This is evident from the careful filed observations and extensive interviews. This initial attitude ultimately finds alignment with their behavior to break lockdown. Why? Reviewing the literature on public administration, government's credibility and accountability and citizens’ evaluation (Dwivedi et al., 2013; Rouban, 2008; Shareef et al., 2019/2012), it is inevitably a common and traditional tendency of the majority of the public to violate (either silently or openly) any government's voluntary instructions which do not comply with their economic, social, emotional, and behavioral requirements. This is the overarching sentiment of this segment of the public who typically perceive any government plan as biased with hidden motives and, thus, their opponent (Berry et al., 2010; Brace et al., 2002). They do not find accountability, transparency, and responsibility from the government's unpopular voluntary discourses and regulations which is not aligned with their spontaneous willingness and, thus, they find no intrinsic motivation to abide by that voluntary system (Ellis, 2010; Erikson et al., 2006). Consequently, and heuristically, this segment of the public is vigorously inclined to violate the central idea of lockdown by maintaining free mixing with social community and professional bodies. This scenario can be explained fundamentally by the central theme of the theory of planned behavior. The construct subjective norm which is clearly an evaluation of the influence of surroundings dictates human psychology to be motivated to pursue an attitude leading to behavioral intention. Since the majority of the public do not comply willingly with a government's sudden and voluntary regulation to follow stringently that which runs counter to their self-interest, they feel the impact of subject norms negatively. This occurs because they do not perceive government's regulation and enforcement favorably in the absence of a transparent and credible image of government. Looking at the balance theory, it is explicitly argued rhetorically that human behavior can be greatly influenced by endorsement of a celebrity if that person has a positive image and acceptability among the majority of people. It means, if a person likes a celebrity, that person is potentially inclined to like to follow that celebrity's suggestion and instructions. As a result, even if his/her initial attitude was non-compliant; finally, the person willingly exhibits favorable behavior by aligning the initial non-compliant attitude favorably. From the perspective of the three parameters of sustainability, this perception of non-acceptance of source derogation and their arguments for the lockdown system corresponds to the economic and social dimensions. From interviews, it was abundantly clear that the majority of the public do not have confidence and trust that the government can and will take sufficient responsibility for an indefinite period to understand, compensate, and support their financial requirements. This argument is strongly supported by literature on the public's opinion on the trustworthiness of government (Erikson et al., 2006). Citizens also do not find any tangible evidence and logical justification from their self-centered perception that government is aware of citizens’ personal belongings and feeling of responsibility for their relatives (particularly elderly parents), friends, and colleagues and many other social members. Consequently, the public find that, if they abide by the idea of a lockdown system, their regular social responsibility is hampered severely. Referring to the response of the public in the interviews, it is particularly clear that citizens perceive government and their associated organizations as inconsiderate of their own economic and social needs. Public administration literature on citizens’ trust of government performance supports this finding (Ellis, 2010; Shareef et al., 2019). In addition to the tangible appeal of upholding economic and social benefits, citizens can also be characterized with another dimension for sustainability of any system, i.e., the behavioral aspect. Evaluation of the source derogation and argument (SDA) has a significant correlation to the behavioral traits of individuals (Alvaro and Crano, 1996; Cohen et al., 1987; Moskowitz, 2005). Social psychologists have provided a strong argument from their analysis that different people have a different mental status in giving of their trust on some social issue (Jones, 1996; Moskowitz, 2005; Shareef et al., 2020b). Evaluating this phenomenon in the context of the social psychological aspect of attribution theory (Heider, 1958), it is quite evident that people's ability, scope and tendency to understand, explain and perceive any social issue and source derogation varies significantly, and it is potentially a unique trait of human behavior which differs from person to person (Kassin et al., 2019). External attribution also frequently termed as situational attribution, is a distinct trait of the human being to portray and evaluate external events favorably or unfavorably, trustworthy or not trustworthy from a truly personal choice in considering the situation. This interpretation is related to personal behavior. Internal attribution referred to as dispositional attribution distinctively reflects evaluation of any events from personal attitude (Heider, 1958). Evaluating this combined phenomenon of human behavior postulates that sometimes human beings interpret a social system and its sustainability from a truly and distinctively personal behavioral motive (BM) and characteristics, irrespective of economic, social, and environmental dimensions and requirements. This behavioral motive (BM) dimension should be considered for the sustainability of any social system with the conceptual definition of an affective urge to uphold any social system without having regard to the economic, social, and environmental requirements necessary for the sustainability of that system. 5.2 Tangible need and deficiency (TND) During this extraordinary situation, all social entities have certain basic needs in respect to the deficiencies they are facing. According to motivation principles and social psychology, while experiencing any deficiencies, several needs surge and emerge in human minds to recover and fulfill those deficiencies (Maslow, 1943; McClelland, 1988). Unless those deficiencies are met, human beings feel discomfort, develop the urge to gain something to fulfill the deficiencies, and relentlessly strive with challenge and motivation to be satisfied (Jones, 1996). This is the driving force of human minds to fight for existence. Social psychologists and organizational behaviorists acknowledge this generic behavior and tendency of human beings and identify their eagerness and efforts (Kassin et al., 2019). Largely identical and similar behavioral patterns are observed among the public during the present lockdown system resulting in a system being executed to confine them in their residential location and keep them detached from all economic activities and social interactions within community for an indefinite period. During this period, from the interviews, it is quite evident that citizens are visibly and logically stressed because of deficiencies in several areas and are facing a serious challenge to fulfill those tangible and realistic needs; for instance, the need for food, drugs, money, future security, job, family responsibility, duties to parents, friends, colleagues, and community, etc. This study examines the external behavior of the public during this period and their temptation to violate the lockdown system due to an inner realization of tangible need and deficiency (TND). This concept can be illustrated with the following definition: As the reason for their negative response to the framework of the lockdown of social isolation, it is the generic perception of the public, through the cognitive realization and understanding about their requirements, that they need to flout the law of lockdown in order to meet visible and realistic deficiencies. This emotional response is well supported and referenced by the motivation theory principle of Maslow's hierarchy of needs (1943). From the outcome of interviews of the three categories of stakeholders in this new study, it is unanimously identified that the public are mostly conscious, concerned, and worried about their future survivability in terms of family income; so they are enthusiastically and desperately striving to violate lockdown to search for an alternative source of income for their present existence. This argument is strongly supported by the first phase of motivation ‘physiological need” of Maslow's hierarchy of need theory. Other content theories of motivation, like McClelland need theory (1988) also provides strong support to explain the public's present feelings of need and deficiency and their action of behavior to fulfill those needs by violating or at least undermining the system of lockdown which excludes them from all economic and social activities for an uncertain period. Now looking at the cognitive dissonance theory to analyze this aggressive behavior of the public to break the lockdown and create a situation of apparently unnecessary social unrest, it can be explained by the fact that human beings, when perceiving misalignment between their attitude and behavior, feel uncomfortable. It may push them to take decisive action. From this attitude, since they relentlessly experience the deficiencies of several visible and essential household materials, they inevitably and eagerly think of ways of meeting these issues immediately in this uncertain situation. On the other hand, they are confined in their houses which hinders any efforts or possibility of meeting those requirements. This is a serious and profoundly felt challenge which suddenly pushes them to break the system of lockdown, go outside, and gain some benefits which can at least partially fulfill certain needs. This way, the public fulfill, or at least try to achieve and feel a certain level of comfort by aligning their behavior and attitude in the same line. Obviously, this challenge is closely associated with social and economic dimensions of sustainability. As per the recommendation and assertion of theorists of social sustainability, out of three parameters, economic and social dimensions are two important aspects of any social system to be met for its existence (the environmental dimension is not relevant in this situation). From the aforementioned arguments, conceptual definitions, and illustrated paradigm, it is explainable that tangible need and deficiency (TND) is the certain realistic demand and effort to fulfill several visible requirements which are closely intertwined with economic existence and social responsibility. Without meeting economic needs (for example, earning, job security, future scopes, present shortage, household essentials) and social needs (like responsibility to other close members, parents, brothers and sisters, friends, colleagues, and community members), the sustainability principle argues that no social systems like lockdown can be effective, successful, and achieve its desired goal. 5.3 Intangible desire and expectancy (IDE) Outside of visibly realized deficiencies, human beings always have several hidden mental desires and expectancies which may surge during this period and this is evident from the interviews. Closely deliberating over the reasons identified by the three stakeholders, it is quite interesting and noteworthy that the sought causes are divisible sharply based on visible realistic items and some mental items denoting heart-felt desire. As such, these requirements can be differentiated into two categories, tangible and intangible. The previous one is already described. Now the latter issue, intangible desire and expectancy (IDE), is examined. Psychologists have long sought to analyze human minds with regard to the fact that, while facing any embargo or the forbidding by some external forces to perform something, internal desire grows and eventually propagates to do that apparently banned action (Tetlock et al., 2000; Wegner and Bargh, 1998). Human psychologists have revealed that it is a common human desire to do something which is not allowed in some specific situations (Wegner, 1994; Zimbardo et al., 1993). A strong desire grows uncontrollably, flourishes, and jumps into action to break that imposed barrier to get the mental relief of freedom and capability to perform following an implied expectation (Trivers, 1971). Organizational theorists also provide rationales and justifications for this human behavior, shedding light on causes pursuing intrinsic motivation (Alvaro and Crano, 1996; Chen and Lin, 2015; Cohen et al., 1987; Moskowitz, 2005; Yin et al., 2015). The Expectancy Theory of Victor Vroom (Vroom, 1964) asserted that, if a human being finds a mental assurance of achieving any recognition by engagement with full efforts, they are enthusiastic to do this with motivation. From interviews, it is revealed that several reasons pointed out by the stakeholders are not rooted in visible logic; some are clearly an internal expression of fulfilling mental need. This is quite logical. Human beings cannot be expected to behave without emotional desire when they face a shortage of such issues. Particularly, during a crisis, when mentally someone feels that they cannot fulfill some desires right now due to many obstructions imposed by society, those desires erupt more aggressively to be fulfilled by violating a social lockdown which has substantial theoretical support from human psychology (Aronson et al., 1995; Fiske and Tetlock, 1997). People miss their routine life, their friends, colleagues, their professional attachment manifested by social capital, which are forbidden now for an unspecified time (Gharehgozli et al., 2020; Martin et al., 2020). They are also not permitted to enjoy their morning walk, evening jogging under the open sky, walking along the road in their familiar community, taking the air without any embargo. Consequently, as a human being, they feel the earnest desire and expectancy to fulfill emotional feelings by forgetting the necessity for social lockdown which has been imposed on them for the sake of their lives. This human behavior can be described and portrayed by the theory of intangible desire and expectancy (IDE) and the paradigm can be defined as: As the reason for the response against the framework of lockdown in social isolation, it is the generic perception of the public through the affective feelings about their requirements to fulfill emotional desire and expectations. This psychological behavior is relevant to the theory of suicidal behavior (Joiner, 2005). Human beings prefer any destructive situation and embrace that situation eventually when they find no other route to escape from any psychological status perceived as stressful, boring, and provocative. Drawing on the analogy of human behavior to change a life status destructively during lockdown period, parallels in human behavior to change the status of confinement is observed. They find some clues and justified reasons to violate the lockdown which restricts them from socialization with their community. Confinement is distressful and breaking the theme of lockdown is more conducive to fulfilling their intrinsic and extrinsic urge and necessity. This provocation and indulgence is a confluence of the respondents’ insight and foresight speculations and future anticipation. In favor of the aforementioned arguments, Maslow's hierarchy of need theory and cognitive dissonance theory can provide background support. As per our previous explanations, people always have a tendency to keep their behavior in the same line of desire derived from inclined attitude. During lockdown, as already illustrated, the public have many internal desires to accomplish which suddenly remain unfulfilled due to isolation from community and confinement in their residential premises. An attitude developed reflecting those desires and beliefs should have an urge to be aligned with behavior to violate lockdown in order to alleviate dissonance (Supported by Leon Festinger, 1962). Maslow's theory (1943.) can also provide an understanding of people's desire for emotional relationships within society and self-esteem as the third and fourth stage demand for intrinsic motivation. In comparison to the presumable requirements for sustainability, fulfillment of social aspect (social dimension) is surely an overarching concern in this context. Plausibly, this intangible emotional desire is not considered to uphold any issue related to economic and environmental aspects. This emotional desire and expectation, as per the studies of human psychology, is certainly very unique and distinct and may be congruent with personal human traits (Daly and Wilson, 1999; Weinstein, 1994; Wegner and Bargh, 1998). Arguably, the feeling and urge of intangible desire and expectation (IDE) varies from person to person and is an essential demand for sustainability to conserve one's own behavioral emotion. However, following the conceptual definition of this theory of intangible desire and expectation (IDE), there is another dimension which is a prerequisite for sustainability of any social system other than the social dimension. As explained before, this emotion is truly and distinctively portraying one's own behavioral characteristics and is a prerequisite condition for sustainability, irrespective of economic, social, and environmental dimensions and requirements. It is an appeal to reduce and get relief from unbearable stress and fulfill intrinsic demand for socialization. This dimension for sustainability can be termed, in this context, as behavioral motive (BM) which is defined previously. 5.4 Evaluation of benefit and loss (UBL) This is a pragmatic evaluation by the human being, and as a rational person, it is a common habit to evaluate both benefits and losses before taking a final decision (Daly and Wilson, 1999; Frey and Osborne, 2017; Trivers, 1971; Tversky and Kahneman, 1992). Marketing, psychology, and organizational researchers (Ajzen, 1991; Alvaro and Crano, 1996; Aronson et al., 1995; Cohen et al., 1987; Dulaimi et al., 2003; Moskowitz, 2005) have analyzed human behavior for many years and have decided that any decision in favor of any actions depends on several interconnected issues which is relevant to its relative favorable merit. Generally, and relatively, any social action is intertwined with several benefits and losses in the context of specific situations, surroundings, time, and associations (Thornton et al., 2011). After analyzing and deliberating over the pros and cons of any actions, depending on the possibility of overall benefits to be gained against risks of losses, people finalize their decision whether they will perform that action or not. Particularly, social economists have asserted that for any incidents associated with economic benefits and losses, people's tendency is to evaluate its relative merits and then take a final decision, either favorable or not (Cohen et al., 1987). The presently imposed and implemented lockdown, as a social system, has several issues which are closely connected with economic activities. While keeping the public apart from their jobs, earnings, and other economic transactions, isolating them from all social interactions, community gatherings, and prohibiting them from performing routine responsibilities for people dependent on them (like parents, relatives, friends, colleagues, and other community members), there exist enormous multidimensional risks, gains, benefits, losses, and problems. They might lose their jobs. The possibility of income opportunities and scopes might be reduced. Many financial transactions may not be performed on time. The opportunity for getting newer jobs in the future might shrink. They cannot perform many duties and responsibilities. On the other hand, the Covid-19 infection is dangerous and life threatening. It is a serious concern not only for oneself, but also for other family members. Following government regulation and maintaining social isolation is a social and civic responsibility and matter of accountability. Giving time to family members, while getting scope in this present scenario, is also a positive aspect. So, there are many issues which have positive and negative aspects. A person, thoroughly, cognitively and affectively, evaluates and takes final decisions based on, and in line with, their own mental ability. Based on the final outcome of a decision, people behave accordingly. This paradigm, following the final reasons revealed by evaluation, can be illustrated by the term evaluation of benefit and loss (UBL). As the reason for a response to the framework of lockdown of social isolation, it is the judgment of the public through the cognitive evaluation about relative gain and loss from the decided action. This phenomenon can be well explained in the light of prospect theory. This theory suggests that before deciding to take any action, it is common human behavior to compare, contrast, and finally evaluate its relative merit by judging the possibility of potential losses and potential gains. It is a relative judgment. Gains and losses might not be purely financial; it may have other social forms. However, a final decision depends on weighing all the issues in terms of opportunities and risks and, thus, benefits and losses in light of comprehending all the issues combined. A rational decision-making model also certifies this behavior of contrasting potential gains achieved and potential losses incurred (Bazerman, 2005; Brunsson, 1989; Cohen et al., 1987). Williamson (1987) in transaction cost analysis advocated for this human behavior in order to decide for any action of behavior under a specific situation. For the sustainability of any social system, for instance the lockdown system, this paradigm of evaluation of benefit and loss (UBL) is directly related to economic dimensions. Without meeting economic desire and basic requirements, any social system cannot be sustainable. However, this paradigm also has an urge to uphold the social dimension. Evaluation of benefit and loss (UBL) also has some issues relating to social duties for those outside one's own family, social responsibility for those members who are living with him/her, and for the general community through isolation and lockdown, and this paradigm is essential in protecting the social dimension of sustainability. Therefore, this theoretical construct, evaluation of benefit and loss (UBL) has both economic and social dimensions of sustainability. Finally, after integrating the aforementioned paradigms, a comprehensive grounded theory for adaptability with any social system during a crisis period, for instance, a lockdown violation scenario as a social phenomenon can be proposed: Source Derogation and Argument (SDA), Tangible Need and Deficiency (TND), Intangible Desire and Expectancy (IDE), and Evaluation of Benefit and Loss (UBL) are the primary and fundamental reasons of the response and consequences of human behavior in a certain crisis of a social incident which causes the public to violate the social system (for instance the newly imposed government regulation of lockdown by maintaining social isolation). This grounded theory can be regarded as the paradigm for lockdown violation (PLV). 6 Theoretical and managerial implications This study has enormous potential to contribute in both the areas of academic knowledge and practitioners’ understanding. It has significant potential to enrich behavioral and psychological studies and literature on sustainability. At the same time, government, public organizations such as law enforcing departments, private donors, NGOs, international organizations like WHO, and the United Nations can avail of readymade and up-to-date views from citizens about their reluctant behavior to abide by government regulations for social isolation, termed here as lockdown during a crisis period. Since this study has collected, accumulated, and analyzed views, doctrines, and suggestions of different active and passive members of the lockdown system, it can provide a generic and generalized view of human behavior during a crisis moment with regard to following government regulations to protect themselves and the surrounding society. In terms of sustainability, this research has potential scope to contribute to the understanding of the root causes of the ineffectiveness of lockdown — a system of social isolation. From interviews of several stakeholders, this study has identified some definite and structured reasons about the partial collapse and failure of social restrictions to ensure the desired and targeted goal. Practitioners such as government, law enforcing agencies, and medical professionals can garner deep insight from these findings as to the reasons for breaking lockdown. Behavioral scientists and social psychologists may find this study interesting as the research reveals several hidden and unfocused areas of background information on not obeying social isolation. Therefore, the paradigm for lockdown violation (PLV) is an excellent model to be explored by both academics and practitioners. The outcome of this study can also enrich the present literature on human behavior which it explored in its journey to analyze the public's motives, attitudes, and perceptions. The public's external behavior in response to lockdown and social isolation and their thought process with regard to violation can provide some new avenues of study about human behavior during crisis moments. Behavioral scientists can understand better the public's process of evaluation about the authenticity of the source and their arguments. While experiencing any need and deficiency, they may observe how people strive to meet those requirements overlooking and undermining potential risks of transmission of a deadly virus. This may bring new challenges to the behavioral scientists to motivate, control, and manage human beings for the sake of society by generating behavioral discourses. This kind of generalized crisis moment and the public's global behavior may definitely provide a new reference for behavioral study. Social incidents, interactions, social systems in association with human motives are also contrasted, explained, and revealed, and, thus, this study has made a substantial input into the understanding of social psychology, particularly in the crisis period. This exemplary study has pointed out that the human motivation to violate rules in any dangerous situation depends not only on tangible benefits but also intangible desires and expectations which arise from affective dissonance. It means focusing on cognitive dissonance theory, feelings of discomfort do not generate just from cognitive dissonance but may also flourish from the perception of affective and emotional dissonance. The paradigm for lockdown violation (PLV) and the grounded theory may motivate academics to conduct further research into human psychology and its responses in crisis moments. Lockdown is a social system. However, active social members of the public may not find enough enthusiasm to abide by the government's instructions and regulations. As such, the effectiveness of the lockdown system is put under question. Therefore, the outcome of this research can benefit the literature on sustainability by helping to understand the parameters to be upheld for sustainability of any social system. From sustainability literature, generally it is recommended that any system must be sustainable to fulfill its targeted goal if it can meet requirements from three areas, environmental quality, social responsibility, and economic demand. However, this study, from the analysis of human psychology and behavior during a crisis, has suggested that upholding benefits from those three dimensions alone is not enough for sustainability of any social system during crisis periods. Practitioners can get certain constructive ideas from this finding. In any critical incidents where masses of people are connected, sustainability cannot be ensured without considering and reflecting the human mind. And in this scenario, both cognitive and affective attitudes are equally important. Social isolation is currently recognized as the only effective solution for this pandemic; however, citizens of different countries, irrespective of being developed or developing, are not aligned with government regulations to follow this social isolation for long periods. Governments of different countries are struggling to maintain social isolation. As such, this study can provide a very pragmatic understanding about citizens’ behavior during any crisis moment. Practitioners and policy makers should understand that feelings of insecurity and the perception of uncertainty for their future, and a perceived unreliability around a government's responsibility, cause citizens not to obey and abide by social isolation regulations. The perception of government administrators is that public behavior is very irregular, undisciplined, and that they should be regulated (supported by theory X, McGregor, 1960). During some specific situations when they face challenges to continue their job, income opportunities, or perform social responsibilities due to protective rules and regulations of government, they are disinclined to obey those regulations, considering the government as their counterpart, not allies (Ellis, 2010; Erikson et al., 2006). Particularly, while facing hardship in the short run, the public becomes arrogant and reacts aggressively to any imposed regulations which may restrict their movement without considering the long-term consequences. As a result, many ruling authorities in the world, particularly law enforcing authorities have sometimes the tendency to follow the hardline without any preplanned attempt to convince them with a softer more even-handed approach. On the other hand, psychologists have also postulated that, if two parties (here government and public) bear mistrust mutually for the opposite party, during any crisis moment, this perception of non-confidence for each other may be aggravated, believing the counterpart's action is from self-interest. While the governments of different countries are imposing lockdown for their citizens, restricting them from community gathering and professional activities, a substantial portion of citizens do not take this regulation seriously. This is an explicit observation and theory for the policy makers which has been derived and proved from this study. This identification is supported by public administration literature (Ellis, 2010; Erikson et al., 2006; Shareef et al., 2019). The human mind is very critical, complex, and it has some internal desires which must be fulfilled for the sustainability of the system, otherwise some people will not be motivated to abide by and execute the norms of the social system, and, ultimately, the social system will fail to fulfill its targeted goal (Bengisu and Nekhili, 2006; Gavilan et al., 2020). This parameter which is essential for the sustainability of the social system is not connected with generalized economic, environmental, and social dimensions, rather it is very much unique and distinctive to individual behavior. Without fulfilling certain intangible desires of the human mind, no social system can be effective and, thus, sustainable. This dimension is referred here as the behavioral motive (BM) dimension. Realistically, for a social system, specifically during an emergency period, this new dimension can be analyzed by sustainability literature. 6.1 Limitations and future research directions As an extraordinary exploratory study, this study has several limitations. It is not developed based on information gathered over the long term, but rather from the recent lockdown system which is a new phenomenon. In future, prospective researchers can collect more information from many countries and experiential settings to expand on these findings. The outcome of this research has proposed a grounded theory rooted on the paradigm for lockdown violation (PLV). This overarching model should be tested through a quantitative study. Since this model is developed based on the investigation in a single and developing country, this is a limitation of this study. It is suggested to test this model in both developed and developing countries to understand generalized human behavior. Future researchers can develop measurements of the four paradigms and constructs of the proposed grounded theory and launch a quantitative positivistic study to verify the validity of the model in any circumstances generalized to relevant crisis events. This theory is neither country nor culture dependent. Cultural experts might have the opinion that cultural attributes should be incorporated as the moderating variables to test their effect on the four paradigms to formulate a response behavior of the public during crisis moments. With definite propositions and quantification of behavioral models as the fourth dimension of sustainability, social scientists can explore the sustainability of the social system. 7 Conclusion In some countries, particularly developed countries, lockdown has ensured exemplified benefits in terms of stopping spread of corona virus; for example, New Zealand, Canada, Germany, The Republic of Ireland, Norway, China, South Korea and Japan. Citizens of these countries have shown more inclination to follow and maintain social distancing and abide by government regulations. As a result, infection and mortality rate is much lower in these countries than many other developed countries like the USA, Italy, Spain, The United Kingdom, France etc. However, at present, worldwide, the most common, frustrating, and popular regulatory word is lockdown, a social system to keep the public confined in their homes or within defined areas for social isolation. For this generation it is a novel, unexpected, phenomenal, and an unimaginable event in the lives of the citizens of the earth. Nevertheless, worldwide, irrespective of classifications such as developed or developing countries, irrespective of country, race, gender, color, or class, a previously unencountered or countenanced disease in the name of Covid-19 is spreading and making the lives of all people worldwide miserable in a unique and distinctly stressful manner (Ellyatt, 2020). Due to its virulent, unbelievable and non-discriminatory capability to spread socially and transmit among communities, Covid-19 has crippled all governments in the world and forced various and unprecedented levels of lockdown socially, nationally, and globally to be introduced (Zhonge et al., 2020). It is forcing governments globally to halt all kinds of economic, social, and individual activities. For an indefinite period, the most powerful governments have helplessly postponed all kinds of routine activities, heretofore considered vitally important for existence. Since this Covid-19 virus is extremely infectious and can be transmitted by any form of close contact, all citizens are advised, instructed, and mandated to be isolated from their jobs, family gatherings, and community activities. Under this context, the public are instructed to maintain lockdown in their homes and to maintain social isolation. So far, medical scientists and WHO have identified that isolation from the community and lockdown is the only feasible solution to counterattack this deadly virus and keep all human beings and society risk free from the transmission of Covid-19. Consequently, governments all over the world have passed and announced recommendations, regulations, punishments, and rewards, advising their citizens to maintain lockdown, the social isolation system (Zhonge et al., 2020). The question is, are the public obeying this lockdown, which, up to the present moment, is the only known solution to help the world, society, community, and individuals to survive? Unfortunately, it is not being strictly followed by the public. Many are not interested or happy to be locked down in their homes for an indefinite period for many reasons; obviously the dominant reasons are related to economic, social, and behavioral issues. Many people are frequently breaking the rule of social isolation, sneaking out to the neighborhood community, and enthusiastically trying to come close to their social friends maintaining the inherent urge for social capital (supported by Erikson et al., 2006). Violating social isolation which can be achieved through lockdown is devastating as it can increase the number of infections in a geometric rate. Due to violation of the central theme of lockdown, in Italy, France, Spain, The United Kingdom, Belgium, and The USA and many others, corona virus transmission is causing a vast surge in the number of deaths worldwide. However, this transmission can only be controlled, and the death rate minimized by maintaining lockdown stringently. The public must abide by government regulations to keep themselves isolated and separate from community interactions. They need to stay at home. However, regrettably, it has been frequently observed that in almost every country, the lockdown system is not completely successful and effective in order to achieve its desired goal. Consequently, corona virus has maintained its momentum and ability to rapidly spread among communities. Every day the transmission rate and death rate in all countries (except countries like China, South Korea, New Zealand, Ireland etc. who were able to impose the lockdown strictly) are increasing at an alarming rate. The most powerful counties are no exception. So far, the only remedy, prescribed by WHO, is to maintain social isolation. Therefore, it is an urgent research question to investigate why people do not feel the urge to abide by a government's regulation when it is relentlessly published by several communication modes that this disease is potentially fatal and people must be locked down to protect themselves and their society from transmission of corona virus. Now closely looking at the four paradigms and analyzing their perspectives, concerns, and guiding principles, it is clearly identifiable that a common notion of cause for this non-compliant behavior among the public is shortage of trust and deficiency of confidence in government to protect their future needs (Coombs, 2020). This is a traditional problem prevailing over time. It is not created from one side alone, either the government or the public. It is an acute issue which exists in our society for many years which is causing mistrust, misunderstanding, and scope to be suspicious of each other's behavior (Brace et al., 2002). On one side, focusing on the literature on public administration (Berry et al., 2010; Brace et al., 2002; Ellis, 2010; Erikson et al., 2006; Shareef et al., 2019), the government is still regarded, in this present era, as the ‘governor’ or ‘administrator’. Government is not interested in being identified as the ‘public service provider (PSP)’. Reviewing literature on the public system and citizens’ trust (Berry et al., 2010; Dwivedi et al., 2013; Erikson et al., 2006;), it is clearly established that citizens can be forced by government to obey any system by punishment, but cannot be forced to obey and maintain any system of regulation if it is imposed by government as a voluntary system. Citizens always have mistrust due to the shortage of confidence about government's motives, hidden mission, and attitudes toward them as the governor. As the ruler, they are considered as the opponent of the public's self-interest and behavior (Ellis, 2010; Erikson et al., 2006). As a result, very rationally, whenever the public find any opportunity, they disobey government regulations in order to look after their own interest and protect personal gain. As the solution based on the grounded theory, for confidence, trust, appropriate communication, and realization from the inner mind, regulations should be communicated with the public by experts; for instance, health experts, medical scientists, social psychologists (supported by Balance theory). Governments should also use celebrities to reach out publicly with two-way communication strategies. These celebrities may give confidence, credibility, and authenticity, and thus, acceptability. Also, it is important to craft the statement of discourse on the regulation regarding lockdown with proper analysis of marketing experts for justified argument and counter logic. Precisely, the government and its associated agencies should strive to create and endorse credibility in favor of the source of derogation and its strength of argument to reflect the public's need and deficiencies. Regulatory statements should also consider the public's intangible mind, for instance their desires and expectations. Regulatory instructions must evaluate different segments’ relative gain and loss in terms of their economic, social and behavioral perspectives. As per the outcome of this study, these strategies are the key to delivering the effective management, control, clarity of public discourse and successful outcomes in any social crisis, with a view to encouraging citizens to abide by sustainable control methods in line with scientific realities and the corresponding advice of civil authorities. CRediT authorship contribution statement Mahmud A. Shareef: Conceptualization, Methodology, Data curation, Investigation, Writing - original draft. Yogesh K. Dwivedi: Conceptualization, Methodology, Supervision, Writing - review & editing. Angela Wright: Conceptualization, Writing - review & editing. Vinod Kumar: Supervision, Writing - review & editing. Sujeet K. Sharma: Conceptualization, Writing - review & editing. Nripendra P Rana: Conceptualization, Writing - review & editing. Mahmud Akhter Shareef is a Professor of the School of Business, North South University, Bangladesh. He was a visiting faculty in DeGroote School of Business, McMaster University. He did PhD in Business Administration from Sprott School of Business, Carleton University, Canada. His research interest is focused on online consumer behavior and virtual organizational reformation. He has published papers addressing consumers’ adoption behavior and quality issues of e-commerce and e-government in different refereed conference proceedings and international journals. He was the recipient of more than 10 academic awards, including three Best Research Paper Awards in the United Kingdom and Canada. Yogesh K. Dwivedi is a Professor of Digital Marketing and Innovation, Founding Director of the Emerging Markets Research centre (EMaRC) and Co-Director of Research at the School of Management, Swansea University, Wales, UK. Professor Dwivedi is also currently leading the International Journal of Information Management as its Editor-in-Chief. His research interests are at the interface of Information Systems (IS) and Marketing, focusing on issues related to consumer adoption and diffusion of emerging digital innovations, digital government, and digital and social media marketing particularly in the context of emerging markets. Professor Dwivedi has published more than 300 articles in a range of leading academic journals and conferences that are widely cited (more than 16 thousand times as per Google Scholar). Professor Dwivedi is an Associate Editor of the Journal of Business Research, European Journal of Marketing, Government Information Quarterly and International Journal of Electronic Government Research, and Senior Editor of the Journal of Electronic Commerce Research. More information about Professor Dwivedi can be found at: http://www.swansea.ac.uk/staff/som/academic-staff/y.k.dwivedi/. Dr Angela Wright, MMIIGrad, MBS, PhD, MCIPD (Academic), MA, D-EduLaw is a Senior Lecturer at Cork Institute of Technology, Ireland - soon to be the Munster Technological University, MTU. She is currently both a Lecturer and Research Supervisor on the Masters Programme (MBA) in the Department of organisation and Professional Development. Her main areas of interest include Education, Tourism and Marketing Enterprise and she holds a masters and PhD in these disciplines. She is the author of several books in this field and continues each year to contribute extensively, through her own and her supervised research, to an extensive range of peer reviewed publications related to Education, Tourism, Marketing, Communications, and Business Management topics. Dr Wright maintains direct links to industry and is regularly asked to consult and contribute to the management & marketing issues of the day. Vinod Kumar is a Professor of Technology and Operations Management of the Sprott School of Business (Director of School, 1995–2005), Carleton University. He received his graduate education from the University of California, Berkeley and the University of Manitoba. Vinod is a well known expert sought in the field of technology and operations management. He has published over 150 papers in refereed journals and proceedings. He has won several Best Paper Awards in prestigious conferences, Scholarly Achievement Award of Carleton University for the academic years 1985–1986 and 1987–1988, and Research Achievement Award for the year 1993 and 2001. Vinod has given invited lectures to professional and academic organizations in Australia, Brazil, China, Iran, and India among others. Dr. Sujeet K. Sharma is an Associate Professor in the ‘Information Systems & Analytics Area’ at IIM Tiruchirappalli. Dr. Sharma has been teaching for more than 20 years in Oman, Bahrain, and India, which has given him a wide experience in teaching and interacting with the students across multiplicity of cultural and ethnic backgrounds. Dr. Sharma has published 30 international journal articles with Australian Business Deans Council (ABDC)/Association of Business Schools (ABS) ranking and indexing in the Scopus database. His ResearchGate score is 25.75 and Google scholar citations stand at 1400 (h-index: 18 and i10-index: 29, Keyword: Sujeet Kumar Sharma). His research articles have appeared in well-known refereed international journals including International Journal of Information Management, Government Information Quarterly, Computers in Human Behavior, and Measurement published by Elsevier; Information Systems Frontiers, and Education and Information Technology by Springer; Behavior and Information Technology, Information Systems Management, Interactive Learning Environments and European Journal of Sports Sciences published by Taylor and Francis and other reputed journals published by Emerald among others. Dr. Sharma also serves as a reviewer to a number of A/A* ranked journals namely International Journal of Information Management, International Journal of Production Research, Technological Forecasting & Social Change, Industrial Management & Data Systems, Information Technology and People, Computers and Educations among others. In addition, Dr. Sharma has presented his research in top tier conferences held in USA, Italy, Malaysia and other countries. Dr. Sharma won the Best Research Award in 2018. Nripendra P. Rana is a Professor in Digital Marketing and the Head of International Business, Marketing and Branding at the School of Management at University of Bradford, UK. His current research interests focus primarily on adoption and diffusion of emerging ICTs, e-commerce, m-commerce, e-government and digital and social media marketing. He has published more than 200 papers in a range of leading academic journals, conference proceedings, books etc. He has co-edited five books on digital and social media marketing, emerging markets and supply and operations management. He has also co-edited special issues, organised tracks, mini-tracks and panels in leading conferences. He is a Chief Editor of International Journal of Electronic Government Research and Associate Editor of International Journal of Information Management. He is a Senior Fellow of the Higher Education Academy (SFHEA) in the UK. He is also a Visiting Scholar at Indian Institute of Management Tiruchirappalli in India ==== Refs References Alvaro E.M. Crano W.D Cognitive responses to minority- or majority-based communications: factors that underlie minority influence Br. J. Soc. Psychol. 35 1 1996 105 121 Ajzen I. The theory of planned behavior Organ. Behav. Hum. Decis. Process. 50 2 1991 179 221 Archibugi D. Filippetti A. Frenz M. The impact of the economic crisis on innovation: evidence from Europe Technol. Forecast. Soc. Change 80 7 2013 1247 1260 Aronson J. Blan ton H. Cooper J. From dissonance to disidentification: selectivity in the self-affirmation process J. Pers. Soc. Psychol. 68 1995 986 996 Bardecki M.J. Participants’ response to the Delphi method: an attitudinal perspective Technol. Forecast. Soc. Change 25 3 1984 281 292 Bazerman M. Judgment in Managerial Decision-Making 6th ed. 2005 John Wiley & Sons New York Bengisu M. Nekhili R. Forecasting emerging technologies with the aid of science and technology databases Technol. Forecast. Soc. Change 73 7 2006 835 844 Berry W.D. Fording R.C. Ringquist E.J. Hanson R.L. Klarner C.E. Measuring citizen and government ideology in the U.S. States: a Re-appraisal State Politics Policy Q. 10 2 2010 117 135 Bowonder B. Linstone H.A. Notes on the Bhopal accident: risk analysis and multiple perspectives Technol Forecast Soc Change 32 2 1987 183 202 Brace P.R. Sims-Butler K. Arceneaux K. Johnson M. Public opinion in the American States: new perspectives using national survey data Am J Pol Sci 46 1 2002 173 189 Brunsson N. The Organization of Hypocrisy: Talk, Decisions and Actions in Organizations 1989 John Wiley & Sons Chich-ester, UK Bunker D. Who do you trust? The digital destruction of shared situational awareness and the COVID-19 infodemic Int. J. Inf. Manage 55 2020 102201 Cairns J Jr Will the real sustainability concept please stand up? Ethics Sci. Environ. Polit. 4 1 2004 49 52 Camerer C., E. An experimental test of several generalized utility theories J. Risk Uncertain. 2 1989 61 104 Chen S.-.C. Lin C.-.P. The impact of customer experience and perceived value on sustainable social relationship in blogs: an empirical study Technol. Forecast. Soc. Change 96 2015 40 50 Cohen M. Jaffray J.-.Y. Said T. Experimental comparison of individual behavior under risk and under uncertainty for gains and for losses Organ. Behav. Hum. Decis. Process. 39 1987 1 22 Coombs C. Will COVID-19 be the tipping point for the intelligent automation of work? A review of the debate and implications for research Int. J. Inf. Manage 55 2020 102182 Daly M. Wilson M. Human evolutionary psychology and animal behavior Anim. Behav. 57 1999 509 519 10196040 Diesendorf M. Sustainability and sustainable development Dunphy D Benveniste J Griffiths A Sutton P Sustainability: the Corporate Challenge of the 21st Century 2000 Allen & Unwin Sydney 19 37 WHO, 2020. https://www.weforum.org/agenda/2020/03/todays-who-briefing-eaa3d34289/. Donaldson L. A rational basis for criticisms of organizational economics: a reply to Barney Acad. Manag. Rev. 15 1990 394 401 Doppelt B. Leading Change Toward sustainability. A change-Management Guide For business, Government and Civil Society 2003 Greenleaf Publishing, USA Sheffield Dresner S. The Principles of Sustainability 1st ed. 2002 Earthscan Publications Ltd London 200 Dulaimi M.F. Ling F.Y. Bajracharya A. Organizational motivation and inter-organizational interaction in construction innovation in Singapore Constr. Manag. Econ. 21 3 2003 307 318 Dwivedi Y.K. Impact of COVID-19 pandemic on information management research and practice: transforming education, work and life Int. J. Inf. Manage. 55 2020 102211 Dwivedi Y.K. Shareef M.A. Simintiras A.C. Lal B. Weerakkody V. Adoption behavior for Mobile Health (m-Health) service: a cross-country comparison Gov. Inf. Q. 33 1 2016 174 187 Dwivedi Y.K. Shareef M.A. Pandey S.K. Kumar V. Public Administration Reformation: Market Demand from Public Organizations 2013 Routledge, Taylor & Francis New York, NY 10017, USA Ellis Christopher R. Why the new deal still matters: public preferences, elite context, and American Mass Party Change, 1974-2006 J. Elect. Public Opin. Parties 20 1 2010 103 132 Ellyatt H. Italy Could Relax Lockdown Measures Within weeks, But Much of Europe reels As Virus Nears Peak 2020 CNBC April 9, 2020 https://www.cnbc.com/2020/04/09/coronavirus-latest-italy Erikson R.S. Wright G.C. Mclver J.P Public opinion in the states: a quarter century of change and stability Jeffrey E Public Opinion in State Politics 2006 Stanford University Press Cohen. Stanford 229 253 Festinger, L. (1962). Cognitive dissonance. Scientific American. 207 (4): 93–107. Figueiredo, A., Codina, A.D., de Figueiredo, D.C.M.M., Saez M., León, A.C. 2020. Impact of lockdown on COVID-19 incidence and mortality in China: an interrupted time series study. [Submitted]. Bull World Health Organ. E-pub:6 April doi: 10.2471/BLT.20.256701. Fiske A. Tetlock P.E. Taboo trade-offs: reactions to transactions that transgress spheres of justice Polit. Psychol. 18 1997 255 297 Frey C.B. Osborne M.A. The future of employment: how susceptible are jobs to computerisation? Technol. Forecast. Soc. Change 114 2017 254 280 Gavilan D. Fernández-Lores S. Martinez-Navarro G. Vividness of news push notifications and users’ response Technol. Forecast. Soc. Change 161 2020 10.1016/j.techfore.2020.120281 Gharehgozli O. Nayebvali P. Gharehgozli A. Zamanian Z. Impact of COVID-19 on the economic output of the US outbreak's epicenter Econ. Disasters Clim. Change 4 2020 561 573 Heider F. The Psychology of Interpersonal Relations 1958 John Wiley & Sons Joiner T.E. Jr. Why People Die By Suicide 2005 Harvard University Press Cambridge, MA Jones K. Trust as an affective attitude Ethics 107 1 1996 4 25 Kaburlasos V.G. Vrochidou E. Social robots for pedagogical rehabilitation: trends and novel modeling principles Dimitrova M. Wagatsuma H. Cyber-Physical Systems for Social Applications; Advances in Systems Analysis, Software Engineering, and High Performance Computing (ASASEHPC) 2019 IGI Global Hershey, PA, USA 1 21 Kahneman D. Tversky A. Prospect theory: an analysis of decision under risk Econometrica 47 2 1979 263 291 Kassin S. Fein S. Markus H.R. McBain K.A. Williams L. Social Psychology 2019 Australian & New Zealand Edition. Cengage AU Katafuchi Y. Kurita K. Managi S. COVID-19 with stigma: theory and evidence from mobility data Econ. Disasters Clim. Change 2020 1 25 Kirkby J. O'Keefe P. Timberlake L The Earthscan reader in Sustainable Development 1st ed. 1995 Earthscan Publications Ltd London 371 Kumar S. Managi S. Does stringency of lockdown affect air quality? Evidence from Indian cities Econ. Disasters Clim. Change 2020 10.1007/s41885-020-00072-1 Kumar V. Kumar U. Shareef M.A. Electronic government adoption: controlling effects of demographic factors Amity Glob. Bus. Rev. 6 1 2011 30 40 Kurita K. Managi S. COVID-19 and stigma: Evolution of Self-restraint behavior 2020 University Library of Munich Germany MPRA Paper No. 103495 Langer M.E. Scho¨n A. Enhancing Corporate sustainability. A Framework-Based Evaluation Tools For Sustainable Development 2003 Forschungsschwerpunkt Nachhaltigkeit und Umweltmanagement, Wirtschaftsuniversita¨t Wien Vienna Lau H. Khosrawipour V. Kocbach P. Mikolajczyk A. Schubert J. Bania J. Khosrawipour T. The positive impact of lockdown in Wuhan on containing the COVID-19 outbreak in China J. Travel. Med. 2020 10.1093/jtm/taaa037 Lozano R. Envisioning sustainability three-dimensionally J. Clean. Prod. 16 17 2008 1838 1846 Luciano E.M. Information management hits and misses in the COVID19 emergency in Brazil Int. J. Inf. Manage 55 2020 102194 Martin A. Markhvida M. Hallegatte S. Walsh B. Socio-economic impacts of COVID-19 on household consumption and poverty Econ. Disasters Clim. Change 4 2020 453 479 Maslow A.H. A theory of human motivation Psychol. Rev. 50 4 1943 370 396 McClelland D. Human Motivation 1988 Cambridge University Press McGregor D. The Human Side of Enterprise 1960 New York Montabon F.L. Pagell M. Wu Z. Making sustainability sustainable J. Supply Chain Manag. 52 2016 11 27 Moon M.D. Triangulation: a method to increase validity, reliability, and legitimation in clinical research J. Emerg. Nurs. 45 1 2019 103 105 30616761 Moskowitz G.B. Social cognition: Understanding self and Others 2005 Guilford Press New York, NY Nakamura H. Managi S. Airport risk of importation and exportation of the COVID-19 pandemic Transp. Policy (Oxf) 96 2020 40 47 32834679 Neumann J.von Morgenstern O. Theory of Games and Economic Behavior 3rd ed. 1953 Princeton University Press Princeton, NJ Pachidis T. Vrochidou E. Kaburlasos V.G. Kostova S. Bonković M. Papić V. Social robotics in education: state-of-the-art and directions Advances in Service and Industrial Robotics 67 2019 Springer Cham, The Netherlands 689 700 Patton M. Qualitative Research and Evaluation Methods 3rd ed. 2002 Sage Publications Thousand Oaks, CA Porter M.E. Competitive Advantage of Nations 1998 The Free Press New York, NY Rees W.E. An ecological economics perspective on sustainability and prospects for ending poverty Popul. Environ. 24 1 2002 15 46 Rey-Martí A. Ribeiro-Soriano D. Palacios-Marqués D. A bibliometric analysis of social entrepreneurship J. Bus. Res. 69 2016 1651 1655 Ribeiro-Soriano D. Salvador R.- D. Linking globalization of entrepreneurship in small organizations Small Bus. Econ. 32 3 2009 233 239 Robinson J. Tinker J. Reconciling ecological, economic, and social imperatives Schnurr J. Holtz S. The Cornerstone of development: Integrating environmental, social, and Economic Policies 1998 IDRC-International Development Research Center, and Lewis Publishers Ottawa 9 43 Roseland M. Sustainable community development: integrating environmental, economic, and social objectives Prog. Plann. 54 2 2000 73 132 Rosner W.J. Mental models for sustainability J. Clean. Prod. 3 1–2 1995 107 121 Rouban L. Reform without doctrine: public management in France Int. J. Public Sect. Manag. 2 2008 133 149 Schept J. A lockdown facility … with the feel of a small, private college: liberal politics, jail expansion, and the carceral habitus Theoretical Criminology 17 1 2013 71 88 Senge P.M. The Fifth discipline. The art & Practice of the Learning Organization 1999 Random House Business Books London Shareef M.A. Dwivedi Y.K. Kumar V. Mahmud R. Hughes D.L. Kizgin H. The inherent tensions within sustainable supply chains: a case study from Bangladesh Product. Plan. Control 31 11–12 2020 932 949 Shareef M.A. Kapoor K.K. Mukerji B. Dwivedi R. Dwivedi Y.K. Group behavior in social media: antecedents of initial trust formation Comput. Human. Behav. 105 2020 106225 Shareef M.A. Raman R. Baabdullah A.M. Mahmud R. Ahmed J.U. Kabir M.H. Kumar V. Kumar U. Akram M.S. Kabir A. Mukerji B. Public service reformation: relationship building by mobile technology Int. J. Inf. Manage 49 2019 217 227 Shareef M.A. Archer N. Dwivedi Y.K. Examining adoption behavior of mobile government J. Comput. Inf. Syst. 52 2 2013 39 49 Shareef M.A. Archer N. Dwivedi Y.K. Mishra A. Pandey S.K. Transformational Government through eGov: Socioeconomic, Cultural, and Technological Issues 2012 Emerald Group Publishing Limited UK Shettleworth S.J. Cognition, evolution, and Behavior 1998 Oxford University Press Oxford Tetlock P.E. Kristel O.V. Elson B. Green M. Lerner J. The psychology of the unthinkable: taboo trade-offs, forbidden base rates, and heretical counterfactuals J. Pers. Soc. Psychol. 78 2000 853 870 10821194 The Guardian China's Coronavirus Lockdown strategy: Brutal But Effective 2020 https://www.theguardian.com/world/2020/mar/19/china March 19, 2020 Thornton P.H. Ribeiro-Soriano D. Urbano D. Socio-cultural factors and entrepreneurial activity: an overview International Small Bus. J. 29 2 2011 105 118 Trivers R.L. The evolution of reciprocal altruism Q. Rev. Biol. 46 1971 35 57 Tversky A. Kahneman D. Advances in prospect theory: cumulative representation of uncertainty J. Risk. Uncertain. 5 4 1992 297 323 Venkatesh V. Impacts of COVID-19: a research agenda to support people in their fight Int J. Inf. Manage 55 2020 102197 Vroom V.H. Work and Motivation 1964 John Wiley and Sons New York, NY Wang C. Horby P.W. Hayden F.G. Gao G.F. A novel coronavirus outbreak of global health concern Lancet 395 2020 470 473 (2020) 31986257 Wang L. Wu J.T. Characterizing the dynamics underlying global spread of epidemics Nat. Commun. 9 2018 1 11 (2018) 29317637 Weinstein A. Market Segmentation: Using Demographics, Psychographics, and Other Niche Marketing Techniques to Predict and Model Customer Behavior 1994 Probus Publishing Co. Chicago, IL Wegner D. Bargh J. Control and automaticity in social life Gilbert D.T. Fiske S.T. Lindzey G. Handbook of Social Psychology 4th ed. 1998 McGrawHill New York 446 496 Wegner D. Ironic processes of mental control Psychol. Rev. 101 1994 34 52 8121959 WHO 6 Steps Every Country Must Take Now to Prevent Coronavirus deaths: WHO Director-General, Director General 2020 https://www.weforum.org/agenda/2020/03/todays-who-briefing-eaa3d34289/ Williamson O.E. Transaction cost economics J. Econ. Behav. Organ. 8 4 1987 617 625 Wood W. Attitude change: persuasion and social influence Annu. Rev. Psychol. 51 2000 539 570 10751980 Yin F.-.S. Liu M.-.L. Lin C.-.P. Forecasting the continuance intention of social networking sites: assessing privacy risk and usefulness of technology Technol. Forecast. Soc. Change 99 2015 267 272 Yoo S. Managi S. Global Mortality Benefits of COVID-19 Action Technol. Forecast. Soc. Change 160 2020 120231 Zhong P. Guo S. Chen T. Correlation between travellers departing from Wuhan before the spring festival and subsequent spread of COVID-19 to all provinces in China J. Travel. Med. 2020 10.1093/jtm/taaa036 Zimbardo P.G. LaBerge S. Butler L.D. Psychophysiological consequences of unexplained arousal: a posthypnotic suggestion paradigm J. Abnorm. Psychol. 102 1993 466 473 8408959

0

PMC9753930

NO-CC CODE

2022-12-16 23:26:22

no

Technol Forecast Soc Change. 2021 Apr 5; 165:120531

utf-8

Technol Forecast Soc Change

10.1016/j.techfore.2020.120531

oa_other

==== Front Long Range Plann Long Range Plann Long Range Planning 0024-6301 0024-6301 Elsevier Ltd. S0024-6301(21)00001-7 10.1016/j.lrp.2021.102070 102070 Article Humanizing strategy Nonaka Ikujiro a∗ Takeuchi Hirotaka b a Professor Emeritus, Hitotsubashi University, Japan b Professor of Management Practice, Harvard Business School, USA ∗ Corresponding author. #702, 2-1-2 Hitotsubashi, Chiyoda-ku, Tokyo, 101-8439, Japan. 8 1 2021 8 2021 8 1 2021 54 4 102070102070 27 12 2020 3 1 2021 © 2021 Elsevier Ltd. All rights reserved. 2021 Elsevier Ltd Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. In this article, we apply our latest thinking on knowledge to provide insights on how to reconceptualize strategy to cope with a VUCA world, epitomized recently by COVID-19. We demonstrate that business leaders must draw on phronesis, or practical wisdom, for strategy to become more future-oriented, society-focused, dynamic, and human-centric. Using in-depth case studies, we show how companies will survive in the long run if they start with a moral purpose, and end by offering value to customers, contributing to society, living in harmony with nature, and creating a new and better future. We came up with six practices that enable business leaders to create new and better futures, citing evidences from neuroscience. We conclude that humans should be at the center of strategy, driving future-making with the help of digital-led automation. Reconceptualizing strategy based on this “inside-out” approach, the reward to the company is resilience, longevity, and sustainability. Keywords Inside-out approach to strategy Practical wisdom Future-making Neuroscience Sustainability ==== Body pmcIntroduction The future may look hazy, but humans have the uncanny ability to sniff the wind, sense what lies ahead, and deal with uncertainty in order to make a better future in which to live. More than 50 years ago, Peter Drucker stated that we may not be able to predict the future, but we can “make” the future. His immortal quote has recently been backed up by neuroscience. Research about the brain has discovered that humans have an uncanny ability to see the world not only as it is, but also as it could be. We think “what if” and can, therefore, make our own future using our imagination and instinct. In order to adapt to the new realities of our VUCA (volatile, uncertain, complex, and ambiguous) world, we draw on two research areas close to us. The first is knowledge creation, a subject we have been conducting research for a long time. The second is strategy, a subject we have been teaching at business schools both in the U.S. and Japan. In this article, we apply our latest thinking from our research on knowledge (and now on wisdom) to provide insights on how to reconceptualize strategy for this day and age of unpredictability. We argue that strategy must become more future-oriented, society-focused, dynamic, and human-centric:• The goal of strategy should be about making a better future, not only about generating superior financial returns. • Maximizing the common good should be equally important to strategy as maximizing shareholder value. • Strategists should include society and future as key stakeholders, alongside shareholders, employees, suppliers, customers, and the environment. • Traditional frameworks in strategy (like the Five Forces framework) are basically static in nature, but strategy should be fundamentally dynamic in nature. • Consultants argue that strategy comes from Big Data and analysis, but we believe strategy originates from the heart, sparked by something subjective and personal. • Humans should be at the center of strategy, driving future-making with the help of machines such as AI, IoT, and AR From knowledge to wisdom We postulated in our 1995 book, The Knowledge-Creating Company, that new knowledge is created by converting explicit knowledge to tacit knowledge and vice versa through an interactive process known as SECI (Socialization, Externalization, Combination, Internalization).1 Our research on knowledge creation gave rise to a whole new view of the organization – not as a machine for processing information – but as a living organism. In that book, we emphasized the importance of the lesser-known of the two knowledge types – tacit knowledge — in creating organizational knowledge. Both explicit and tacit knowledge reside in an individual, or to use Peter Drucker's words, in the “head” and “hand” of an individual. When knowledge management became popular in the late 1990s, however, managers and executives in the West tended to accord more value to explicit knowledge. To use a metaphor, explicit knowledge is the tip of the iceberg and tacit knowledge is the much larger base lying at the bottom of the ocean. The expression, “If we only knew how much we know” by Michael Polanyi (1996), a scientist and philosopher, points to this important body of knowledge we call tacit knowledge that lies deep under water invisibly and unconsciously. The advent of the Internet brought with it digital-led automation, such as machine learning, Big Data, data analytics, Cloud computing, Artificial Intelligence (AI), Internet of Things (IoT), Augmented Reality (AR), and much more. Digital-led automation is making data, information, and knowledge – especially explicit knowledge – more abundant, open, and connected:• Knowledge is becoming free, limitless, personalized, and shareable (“Just Tweet it”) • Data, information, and knowledge are becoming indistinguishable • The Internet, combined with social media and mobile technology, is creating a hyper-linked world The buzz over AI has given rise to the belief by some that Singularity – the term Ray Kurzweil coined for the moment AI will supersede humans – will be at hand. The idea is that the computer will become so powerful that it will be able to deal with any challenge we throw at it or solve all of the world's problems. That certainly has not taken place yet. Why? From a knowledge perspective, it is because AI does not have tacit knowledge embedded in it; it is all about explicit knowledge.2 In trying to solve a problem, our tendency is to rely on explicit knowledge and linear causation, even when the messy world we live in defies logical thinking. Practical wisdom atop knowledge To adapt to this messy world, we needed to draw on a third, often forgotten kind of knowledge called phronesis that Aristotle talked about 2400 years or so ago. It is interpreted as experiential knowledge that enables us to make prudent judgments in a timely fashion, and to take actions guided by values, principles, and morals. Among practitioners, it is better known as practical wisdom. In 2019, we published The Wise Company as a sequel to the original book. We have come to realize that knowledge is of no value unless it is put into practice. We listened to an old Japanese proverb, “Knowledge without wisdom is a load of books on the back of an ass.” Thus, our research focus evolved from knowledge to wisdom, which explains why the title of our books evolved from “The Knowledge-Creating Company” to “The Wise Company.” We are reinforcing our research by putting wisdom atop knowledge.3 To understand wisdom better, think of “mother's wisdom.” All of us have a mother and almost all of us gained mother's wisdom by living with her, by watching from behind her, by being scolded by her, and by being told over and over again, to be honest, not to tell a lie or cheat, not to be greedy, not to be a nuisance to others, and more. We remember mother's wisdom because we heard her words repeatedly and ceaselessly. It did not come to us in the form of a formula, a theorem, or manual. Mother's wisdom is appealing because it combines insights, self-sacrifice, love, and social understanding. Mother's wisdom is passed down generations through stories, through narratives, and becomes part of our life through practice, through action. Whereas knowledge becomes obsolete the minute it is created, mother's wisdom endures over time. In Japanese, we go back a generation and refer to mother's wisdom as “grandma's wisdom package.” It is similar to the wisdom that has been passed down generations among the Indigenous communities worldwide. Because of this resilient quality, wisdom serves as a countervailing force in this world of high-velocity change. We also realized that knowledge creation alone is insufficient for knowledge to be useful. Creation can get us new knowledge, but not the wisdom to use it properly. For knowledge to be useful, we concluded that knowledge creation has to be paired up with knowledge practice. Knowledge creation is about acquiring, accessing, accumulating, codifying, and storing knowledge. It can be embodied in new products, new technologies, new systems, new methods, new organizational structures and other forms of innovation. Knowledge practice, on the other hand, is about applying knowledge, putting it to use, disseminating it, and converting it into action. Both mother's wisdom and practical wisdom come to us through practice and are long-term in focus. There is a difference, however, on how scalable mother's wisdom is compared to practical wisdom. Mother's wisdom enriches her children, and other children around her, but practical wisdom applies more broadly, having an impact on individuals, companies, communities, and society at large. Similarly, practical wisdom elevates the knowledge creation and knowledge practice process that starts at the individual level to the societal level over time. In the new book, we made the process more dynamic by adding a third dimension – time – to the original SECI Model and converted it into a three-dimensional model4 called the SECI Spiral Model (see Fig. 1 below). The SECI Spiral, as the name indicates, comes from the spiraling up of the SECI process over time. We did mention that organizational knowledge creation is a spiral process in the original book, but only in passing. We featured the spiral aspect of the SECI process more prominently in the new book to show how knowledge is created, amplified, and practiced over time repeatedly and ceaselessly.Fig. 1 The SECI spiral model. Fig. 1 Although not depicted clearly in Fig. 1, SECI starts with Socialization at the individual level. A dynamic interaction takes place between knowledge creation and knowledge practice within one horizontal round of SECI from Socialization to Internalization. This dynamic interaction also creates a vertical spiral movement to higher levels of ontology – from the individual level to the society level. Knowledge creation is carried out collectively by individuals, teams, and organizations as they interact with the external environment. 5 Fig. 2 Updated SECI model. Fig. 2 Not shown explicitly in Fig. 1 as well, companies make use of “creative routines” – or kata as they are called in Toyota – to adapt dynamically to the changing external environment. At Toyota, kata is defined as “a means for keeping your thoughts and actions in sync with dynamic, unpredictable conditions” (Rother, 2010: 16). This definition suggests that although the conditions surrounding an organization are always changing dynamically, the company can develop creative routines to remain adaptive and future-focused. One of these practices is called “Thinking Two Levels Above,” which forces employees out of their silos and compels them to take a broader view. Just as the view of the landscape changes as the climber goes from the foot of the mountain to the top, this kata encourages employees to imagine what it would be like to view the world higher up in the mountain and to think “what if” about the future (Nonaka and Takeuchi, 2019; Osono et al., 2008). Without going into details, we want to point out three features of the SECI Spiral Model. First, notice how knowledge that is created at the individual level becomes organizational knowledge, then spirals up to the inter-organizational or community level and eventually to the society level. We use the “spinning top” as a metaphor to illustrate the dynamic nature of the knowledge creation/practice process. If the top spins with enough velocity, it will defy the forces of gravity to maintain its balance against shocks and crises coming from the external environment. For companies, that is equivalent to maintaining resilience and being sustainable. The top will fall over if it stops spinning, which, for companies, means death. Second, we also want to draw attention to the vertical arrow in the middle, which represents practical wisdom. Companies need practical wisdom as the driving force to continue spinning so that they can make a contribution to society. As Tadashi Yanai, the CEO of Fast Retailing that operates UNIQLO stores worldwide, pointed out:(S)ociety only recognizes companies, and permits them to survive, if they have some contribution to make from the moment a company is born. It is an instrument of the public. In other words, if companies aren't improving society in some way, society will quickly shut them down (Nonaka and Takeuchi, 2019: 101). Third, look at how the tip of the vertical arrow is pointed a little bit towards the right in Fig. 1. The arrow, however, can be pointed to any direction. In other words, a company has the discretion to choose the direction it wants to move in the future. The future is wide open, 360° to be exact. Practical wisdom serves as the driving force that directs the company to choose a unique future. The future-we-want-to-make should reflect the beliefs and ideals of the company's founder or successor, as we will see in the case of HondaJet below. The HondaJet story Making the future a reality requires a long time. To illustrate, let's look at Honda's foray into business jets. Soichiro Honda, the founder of Honda Motor, was infatuated with airplanes since his childhood days and eventually became a licensed pilot. He made it clear to everyone in the company after his success with motorcycles that he wanted to create a future in which anyone, including himself, would be able to fly an easy-to-handle and low-priced aircraft. That was his dream. As an indication of how much he wanted his vision to become a reality, the first motorcycle the company produced in 1948 was named the Honda Dream, which came with the image of a wing as its product mark on the side of the main body. Every Honda motorcycle to this day carries that wing mark. The founder's dream became a reality in December 2015, when the Honda Aircraft Company in Greensboro, North Carolina delivered the first small business jet to its customer in the U.S. It took the company 29 years to commercialize what industry sources described as the fastest, most energy efficient, highest altitude-reaching, quietest, largest main-cabin, largest cargo-hold, and sexiest-looking jet in its class. The rest is history. In less than two years, the $4.85 million HondaJet became the best-selling small business jet in the U.S.A. Seven generations of CEOs at Honda who succeeded the founder kept Soichiro's dream alive. It was, however, a secret that only the CEO and a handful of engineers inside the company knew. Michimasa Fujino, the CEO of the Honda Aircraft Company who spearheaded the HondaJet project since its launch, was one of them. Soichiro did not even know. To make sure that the project would remain a secret, Fujino spent most of his time developing HondaJet in the U.S. He met Soichiro once, only in passing in the men's room at a research center in Japan, but they did not exchange words. Fujino confided that he regretted not being able to tell Soichiro, who had already been retired by then, that the future the founder had envisioned would soon become a reality. Fujino followed strict orders from successive CEOs not to tell anyone about the project, but he couldn't help feeling remorseful since Soichiro passed away few years after their silent encounter. According to the above story, Soichiro had a dream, while Fujino made it happen. Since we used the words dream and vision interchangeably, we could also say, Soichiro had a vision, while Fujino made it a reality. Vision, as we see it, addresses the question, “What kind of a future do we want to create?” Both dream and vision are about the future, but they are about the future “we” want to create. Thus, they are both subjective and personal. Fujino also confided that the future he wants to create may not be what the founder had in mind. Having lived in the U.S. a long time, the future he wants to create is one in which American executives would be spending less time on air travel, including time at airports, and more time with their families. To go back to Fig. 1, that means the arrow may be pointing in a different direction for Honda Aircraft Company in the U.S. today as compared to the direction that the founder had in mind 70 years ago. Two approaches to strategy Most business schools around the world do not currently teach strategy with the future-we-want-to-make in mind. We would like to change that, but before we dive into doing so, let's make sure we have a common understanding of what is currently being taught in a basic strategy course. At the Harvard Business School, we define strategy as an “integrated set of choices” which positions a firm “in an industry” so as to generate superior financial returns “over the long run.” The foundation of strategy – what we call “blocking and tackling” to use an American football metaphor – is taught in our classroom (and now using Zoom) along the following three modules: 1. External environment (corresponds to “in an industry” above) 2. Internal activities (“integrated set of choices”) 3. Competitive dynamics (“over the long run”) We make use of analytical frameworks that Michael Porter developed in Competitive Strategy, a book published in 1980, to enable students to size up the external environment (Five Forces framework), configure internal activities (Value Chain framework), and anticipate/respond to competitive dynamics (Competitor Analysis framework). If firms do these analyses right, they will most likely be able to come up with strategic positionings that generate superior, sustainable returns (Porter, 1980, Porter and Takeuchi, 2000). This analytical process utilizes what we call an “outside-in” approach to strategy. Strategy is being taught by:• first analyzing what lies outside the firm – industry structure and the competitive field (external environment) • then choosing what to do and what not to do inside the firm (internal activities) The outside-in approach has served us well over the years. The fact that Competitive Strategy is still used as a strategy textbook at hundreds of business schools around the world is a tribute to its intellectual rigor. The more amazing fact is that a revised edition of this book has never been published. Instead, revisions have been made in the form of articles published in the Harvard Business Review. For example, Michael Porter responded to the criticism that thinking only about making our shareholders happy is too narrow by co-authoring “Creating Shared Value.” The CSR article was published eight years before the Business Roundtable in the U.S. issued a statement in 2019 that shareholder value is no longer everything. The message coming from 181 CEOs of the Business Roundtable was very clear – making other stakeholders (customers, employees, suppliers, and communities) happy is equally important. As Tadashi Yanai noted, “Ultimately, the company's goal must be to make people happier …. if you are always chasing money, you'll never catch it.” (Nonaka and Takeuchi, 2019: 101). “Inside-out” approach to strategy We live in a time of flux and fluidity, when the pace of change is more unrelenting that in the past. It's a tough world out there. To survive, strategy has to become more about future-making. To do so, companies have to adopt what we call an “inside-out” approach to strategy. In the HondaJet story, recall how the founder's dream (inside) created a new future aimed at improving society at large (outside). Our inside-out approach to strategy contrasts not only with the outside-in approach mentioned above, but also with prior approaches that focused on the “inside” as consisting of resources (resource-based view) or competencies (competency-based view). In our inside-out approach to strategy, the beliefs and ideals of the founders (or successors in the company) serve as the starting point. It focuses on the origin of the firm, not what it has as its asset. As we see it, strategy is sparked by something subjective and personal, like dreams, gut feel, intuition, insight, imagination, inspiration, and other forms of tacit knowledge. We believe strategy originates from the heart, not from Big Data and analysis, as many consultants contend. Starting from the individual drive to make a difference in the world, strategy moves out to the societal level and into the future, as depicted in the SECI Spiral Model. One of the fundamental questions in strategy is, “Why do firms differ?” The most likely response from the followers of the outside-in approach is, “Because they have different value chains or activity systems.” From our perspective, however, firms differ because they envision different futures. They differ because people in charge of formulating and executing strategy choose a unique future they want to create. Viewed in this way, strategy is about future-making. Strategy is still about making a choice. In addition to “running a different race” and choosing “what not to do,” which we learned from the outside-in approach, strategy is also about choosing a unique positioning with regards to the future we want to make. With practical wisdom as the driving force, the goal is directed outside, namely, to make the future world a better place in which to live. The future must extend beyond the narrow interest of the company. It must be about pursuing the common good. Only then will companies start to think of themselves as social entities that have been charged with a purpose to create lasting benefits for society. Doing so will also recover the founding agenda of the social sciences, which is to improve the human condition. You can hear the cynics groan. That's too idealistic, they may argue. But in The Wise Company, we wrote:(C)ompanies must create a new future in order to survive. Those futures can no longer be extensions of the past; they must be leaps of faith. Leaders cannot be content analyzing situations using empirical data and deductive reasoning; they must also make inductive jumps according to their ideals and dreams. If they aren't idealistic, they simply can't create new futures (Nonaka and Takeuchi, 2019: 23). An endorsement of the inside-out approach to strategy came from Fast Company, a business magazine, which raised the following question to the cynics: “What if an inside-out strategy creates more creative, resilient companies than those following the old outside-in approach?” (Safian, 2014: 74). The payoff of the inside-out approach to strategy, which is widely practiced among Japanese companies, lies in the end result it brings. It comes in the form of resilience, longevity, and sustainability. Japanese companies have been criticized for not being sufficiently capitalistic – not maximizing shareholder value, not laying off employees to reduce costs, not paying compensation that will incentivize top management, and the like in the short run. But history has shown that it pays to have a long-term view. An estimated 20,000 of the 1.24 million companies in Japan have been operating for over 100 years, and about 1200 Japanese companies have been in business for more than 200 years; around 600 companies more than 300 years; and about 30 have survived for more than 500 years. And five companies have lived for more than 1000 years (Funabashi, 2013: 274–275). Six practices to make a better future In the remainder of the article, we examine what is it about humans that drives them to make inductive jumps into the future. We will discuss six practices that humans have carried out to make strategy become more human-centric and future-focused:• Coping with complexity • Adapting dynamically to reality • Embracing “dynamic duality” • Empathizing with others • Narrating stories • Living with nature Wherever possible, we will cite evidences from recent neuroscience research that the body plays a pivotal role in how our brain works. The traditional view in brain science, known as the brain-bound view, took the stance that the brain is solely responsible for our thinking and cognition. By contrast, the “embodied cognition” view has shown that the mind must be understood in how it interacts with the physical body and the outside world (Varela et al., 1993). Coping with complexity The real world, to use the words of David Sax “isn't black and white. It isn't even gray. It is multi-colored.” He goes on to say in The Revenge of Analog that it is “infinitely textured, and emotionally layered. It smells funky and tastes weird, and revels in human imperfection” (Sax, 2016: xviii). Humans do not live in an air-tight closed world that Siri and Alexa live in, says Brandt and Eagle, a composer and a neuroscientist. Our world is open and has “porous borders that leak future. We balance an understanding of our present reality against an imagining of the next. We constantly peer over the fence of today into the vistas of tomorrow” (Brandt and Eagleman, 2017: 250). Our mind is fluid -- free to wonder around, to think laterally and randomly, and to make new combinations. Neuroscientist Christof Koch observes that the human brain is part of a complex integrated system in which the brain, the body, and the world operate as a dynamic system. As a result, our consciousness has “the ability to combine data from different sensors to contemplate and plan a future course of action,” says Koch (Koch, 2012: 129). Adapting dynamically to reality We have known that companies and managers possess the ability to adapt to the environment and even shape that environment, thanks to the work of David Teece on “dynamic capabilities” (Teece, 2009).6 As mentioned briefly earlier, companies rely on creative routines – or kata as referred to in Toyota – to adapt dynamically to the changing reality. Examples of these practices in Toyota include: “Ask why five times,” kanban (the card about components sent along the production line), yokoten (best practices sharing) and jidoka (automation). We now know that the brain, once thought to be static, is always changing. Keith Yamashita of SY Partners, who worked closely with Steve Jobs at Apple, recently reported that the brain changes its shape in several ways: (a) by creating new neural pathways and disrupting old ones, (b) by shrinking specific parts of the brain and growing others, and (c) by lighting up particular brain regions, making it easier and more natural to switch a region depending on contexts.7 In Strategy: A History, Lawrence Freedman, a historian of military strategy, argues that strategy must adapt to change using “soap opera” as a metaphor (Freedman, 2013). In soap operas, new characters constantly come into play and the plot lines unfold in different directions over a series of episodes, oftentimes deflected by chance events. Unlike a three-act play with a definitive ending, each episode is self-contained with no prescribed ending. We cannot predict when, where, and how the central characters and their circumstances change throughout the series. The plot in a soap opera must therefore have a built-in freedom of movement. Freedman contends that strategy is similar to a soap opera as they both have to adapt dynamically to the ever-changing reality. Embracing “dynamic duality” There is a strong propensity in the West to view the world in an “either/or” model. This intellectual tradition can be traced to Descartes and the Cartesian split. To use the A/B analogy, A is pitted against B, resulting in a “A versus B” setup. This intellectual tradition is reflected in the debates over dualism – such as mind versus body, subject versus object, rationality versus empiricism, materialism versus idealism, and much more. In management, it is represented by debates over machines versus humans, analytics versus intuition, economic versus societal value, exploration versus exploitation, egoism versus altruism, etc. Take the classic “body versus mind” dualism as an example. Humans, through practice, can turn it into a “body and mind” duality. A Fortune editorial piece on skiing, says it all:You can plan one or two decisions in advance, but readiness to make an instant adjustment is always required, as everything around you is swiftly and constantly changing. In the flow, you have no time for reflective or analytical thinking. Your body needs to work in concert with the mind absolutely, simultaneously, and unconditionally (Herman, 2015: 30). Thanks to recent findings in neuroscience, you can now kiss the Cartesian split good-bye. We have known all along that humans, compared to other living creatures in the animal kingdom, have an enormous cortex, in particular an outsized prefrontal cortex. We are now finding out that the prefrontal cortex can be trained to pay attention, absorb details, and think clearly through practice. In other words, the brain can be trained to take a divergent phenomenon, as those mentioned above, and convert it to a “both/and” setup through training. Empathizing with others To cultivate wisdom, humans have to empathize with others. Research in both philosophy and neuroscience is backing up our contention that humans have mastered this unique quality and practice it every day. The German philosopher Edmund Husserl (1859–1938), who laid down the foundations of phenomenology, conducted research on how subjectivity can be shared among multiple individuals, known as intersubjectivity (Nonaka and Takeuchi, 2019). He argued that intersubjective experiences stem from empathizing with others, where intentional acts to other subjects are recognized and understood by “putting yourself in someone else's shoes.” Husserl called this mechanism of empathy “pairing.” When two individuals become paired with one another, the narrow egoism that separates the two dissipates, making them feel as if they are directly connected to one another. Recent studies in neuroscience show that empathy – or acting together, cooperating, and caring about others – is facilitated by basic brain functions (Nonaka and Takeuchi. 2019). For example:• Studies on the activities of the medial prefrontal cortex present evidence that the human ability to understand other people's minds occurs at the pre-cognitive level, that is, even before such a consideration is recognized consciously. • Research shows that empathetic feeling triggered by someone's pain involves anterior insula and cingulate cortex. • Discovery of the mirror neuron system in the human brain has opened up the possibility that “mimicking” others might play a critical role in understanding the intersubjective experiences with others. • Scientific evidence suggests that people with active insulae – the body sensation center – tend to have high empathy. Although empathy is facilitated inside the brain, it needs a ba for it to expand beyond the individual level. In Japanese, a ba – which literally translates as place, space, or field – refers to the context in which relationships are forged and human interactions take place. Think of it as a temporal container for creative interaction, where the space, time, and context can be physical, virtual, or cognitive. Empathy lays the foundation for anyone to know anything about others and the world in which we live, involving all five senses of the body, not just the brain. Emphasizing with others is synonymous with sensing and understanding others not just on a shallow intellectual level but on a deep, emotional level. To empathize on a deep level, we need to develop sharp bodily senses and cultivate compassion in our hearts that can capture the feelings of others. When we sense the feelings of others as if we are having such feelings on our own, only then can we create a sense of “us” with others. That's what the practice of “putting yourself in someone else's shoes” means. Narrating stories Stories become a prism through which humans live. Call them narratives, scripts, plots, novels, metaphors, cognitive maps, or soap operas, they define how we live. Stories are the way we understand life and imagine “what could be.” The more vivid the stories we tell of what is possible “out there,” the clearer the future becomes. As James Kerr pointed out in Legacy:… the leader begins to bring the story to visceral life across as many channels as possible. In this way, language becomes the oxygen that sustains belief. In this way, leaders rewrite the future (Kerr, 2013: 154). Business leaders and politicians – as well as journalists, filmmakers, and YouTubers – all know the power of stories, but novelists are a league of its own when it comes to writing epic stories. Business schools and other disciplines teach us that we should empathize with others, but “only literature offers this constant practice in doing so,” according to Gary Morson and Morton Schapiro in Cents and Sensibility. The stories that great novelists write cannot be understood through deductive logic, they contend:Human lives do not just unfold in a purely predictable fashion the way Mars orbits the sun. Contingency, idiosyncrasy and choices – all of which allow for alternatives – play an important role (Morson and Schapiro, 2017: 9–10). In addition to literature, history helps humans to envision “what could be done” in the future. History is the story of human activities that allows us to look back at historical events from the present, interpret and reconstruct the past, and create possible futures. History explains the “why” by describing the causality between the past and the present and the “how” as in “how has this come to be.” Living with nature Putting humans at the center of strategy enables us to live in harmony with nature, which has been around a long, long time – over four billion years. In our view, we homo sapiens (or “wise man” in Latin) have been plagued these days with short-termism and over-planning. Take strategic planning as a case in point. In most companies, it only covers a three- or five-year time span. Worse, a plan seldom touches our heart. Imagine Martin Luther King giving the “I have a plan” speech instead of the “I have a dream” speech. Remember what we pointed out earlier – that strategy originates from our heart. Remember also what we mentioned earlier – brain research has discovered that humans can see the world not only as it is, but also as it could be. We have the ability to run “what if” simulations. Brandt and Eagleman, the composer-neuroscientist pair, observed that the world we live in is “the product of what-ifs built atop one another generation after generation” (Brandt and Eagleman, 2017: 245). As natives of Japan, we are heartened by efforts to live in harmony with nature. For example, Shinto priests at the Ise Grand Shrine have been planting and rebuilding the shrine every 20 years for the last 1300 years to celebrate the renewal and the cyclical quality of nature.8 As this example illustrates, the practice of “oneness of self and nature,” which is one of the most important characteristics of the Japanese intellectual tradition, is still alive and kicking. Only through oneness with nature can we build a lasting future. Reconceptualizing strategy How do we reconceptualize strategy? We recommend making the following three fundamental changes. For companies living in a world of high-velocity change, where discontinuity is the only constant and uncertainty is the only norm, doing nothing is not an option. Not making changes will mean death. To survive, new routines are in order. First of all, companies need to put humans at the center of strategy. Adopt an inside-out approach by letting humans formulate strategy by starting with beliefs, ideals, and intuition. Since we now know from neuroscience that the body and brain are connected, let humans make use of their experiential knowledge to make strategy a way of life for all employees. Let humans be in the driver's seat to deal with unpredictability since the brain has the ability to combine data from different sensors to contemplate a future course of action and handle unexpected and novel situations. Scientific evidence makes it clear that the brain urges humans to empathize with others, so let humans come up with “what-if” options. Let humans be in charge of pursuing the common good to ensure survival. Second, pound the message into everyone's head that strategy should be driven by practical wisdom. Based on our definition earlier, there are two dimensions to practical wisdom: 1) taking actions guided by values, principles, and morals, and 2) making “here and now” judgment calls. On the one hand, strategy must constantly pursue what is good for the company as well as for society. It must constantly pursue the “common good,” making people happier and improving society in some way. On the other hand, strategy needs to be dynamic and agile. Companies have to make judgments knowing that everything is contextual, make decisions knowing that everything is changing, and take actions knowing that everything depends on doing so in a timely fashion. Doing both at the same time – constant and dynamic – is the challenge facing strategy. This tension keeps everyone engaged in ensuring that the top keeps spinning continuously. Third, make future-making the mission of strategy. Because the future is hazy, unpredictable, and leaky, strategists need to rely on story-telling in order to clarify where the company is headed. Narratives become a set of beliefs that bind the organization together, a plot that everyone in the organization can follow. Using narratives, strategy serves as a script for all employees about what the company stands for and what kind of legacy it wants to leave behind. Thinking of the future makes strategy focus on the long term. As the HondaJet story exemplified, making the future you want to create takes time. Thus, strategy is not about creating a three-year or a five-year plan. It is about making your vision, defined as “what kind of a future do we want to create?” into reality. That may take generations. To avoid short-termism, companies need to formulate strategy with the future in mind as well as with nature in mind. If you succeed in making a new future, a better future, the reward to the company is resilience, longevity, and sustainability. Wise capitalism We argued in this article that no company will survive in the long run if it does not start with a moral purpose, and end up offering value to customers, contributing to society, and living in harmony with nature. We close by raising the following questions: (1) Will the old notion of capitalism survive? (2) Is it time for us to evolve into a new form of capitalism in order to adapt to the new realities of our VUCA world, especially in light of what is taking place around the world with the COVID-19 crisis today? In The Wise Company, we pointed out that capitalism, especially as practiced in Wall Street today, typically pits business and society against each other. This divisiveness has given rise to a system based on (a) maximizing shareholder value, quarterly earnings, executive compensation and incentives, (b) reducing costs by laying off employees or moving businesses offshore, (c) putting off concerns of climate change, global warming, and other environmental issues, and (d) strengthening legal compliance through the U.S. Sarbanes-Oxley Act and the like. In July 2019, David Teece invited both of us to an international conference in Edinburgh, Scotland. The topic of the conference was on “reshaping capitalism and the global order.” It was held at the home of Adam Smith, the founder of capitalism, who passed away in 1790. At the end of the two-day conference at Panmure House, which Nonaka attended, the following declaration was signed by over 80 participants:The first declaration of Panmure House urges international leaders to base their policies and decision-making on a set of common principles, as espoused and formulated by Adam Smith, which cherish the required values of an ethically-based liberal democratic system, a moral commitment to the well-being of our communities, and affirm responsibility to protect economic social and political freedoms, and resources, wisely avoid unintentional consequences, follow the rule of law, favor markets and prices as guides to resource allocation and a long-term view of private and private investment to support inclusive economic growth and prosperity for all.9 A month later, in August 2019, 181 CEO members of the Business Roundtable in the U.S. signed a new mission statement, as mentioned earlier. For more than two decades, it explicitly put shareholders first, but its new mission talks first about delivering value to customers, followed by employees, suppliers, communities, and lastly by shareholders. A Fortune article, “America's CEOs Seek a New Purpose for the Corporation” explains why the switch was necessary. It talked about the changing economic, social, and political contexts in the U.S., which moved towards widening economic inequality and deepening mistrust of business in the last two decades. These two events pre-dated the corona virus crisis, which means that we have been concerned about reshaping capitalism and the global order as well as about mistrust and divisiveness in the U.S. even before the outbreak of COVID-19. In 2019, we lived in a world where complexity was already a way of life, but with corona virus, the world has become even more unpredictable. The pandemic has not only aggravated our concerns; it is calling to question whether or not the old form of capitalism will survive. If capitalism were to survive the crisis we are facing today, it has to evolve into a new form of capitalism – call it wise capitalism – which will be based on phronesis. The two of us are hoping to write about wise capitalism next. We wrote “The Wise Leader” ten years ago, The Wise Company last year, so why not Wise Capitalism next? Our vision is to write about a system in which the “new normal” becomes:• Living in harmony with society • Having a social purpose in earning profits • Pursuing the common good as a way of life • Having a moral purpose in running a business • Living in harmony with nature • Evaluating performance based on resilience, longevity, and sustainability Given these values, principles, and morals, we hope to create a narrative for a better future where divisiveness will be replaced with inclusivity, mistrust with empathy, and egoism with altruism. That's too much to expect, you may say. But we need new futures to survive and those futures must be leaps of faith into the vistas of the unknowable tomorrow. To repeat, we need to make inductive jumps according to our ideals and dreams. Who knows, we may have to “humanize” capitalism as well.10 , 11 NONAKA, Ikujiro: Professor Emeritus, Hitotsubashi University Ikujiro Nonaka received his BA in political science from Waseda University in 1958, and MBA and PhD in Business Administration from the University of California, Berkeley in 1968 and 1972, respectively. Professor Nonaka was appointed a Xerox Distinguished Faculty Scholar of the University of California in 1997, Professor Emeritus of Hitotsubashi University in 2006, and University Professor of Waseda University in 2013. Professor Nonaka was the Dean of the Graduate School of Knowledge Science, Japan Advanced Institute of Science and Technology, from 1997 to 2000. Previously he was Professor (1982–95) and Director (1995–98) at the Institute of Business Research, Hitotsubashi University. His earlier academic career included positions at Nanzan University and the National Defense Academy's Faculty of Social Science. Professor Nonaka's primary research interest is to establish and disseminate the theory of knowledge-based management of companies, communities, public administration, and the nations, in order to facilitate ongoing, sustainable knowledge creation and innovation. As part of this work, he has conducted comparative research on leaders and on knowledge-creating processes in companies and organizations, and of leaders, around the world. Accordingly, he is known globally as the ‘guru’ of Knowledge-based Management, having proposed concepts and theories on organizational knowledge creation processes and leadership since the 1980s. His academic works include the SECI Model for the organizational knowledge creating process, the concept of Ba and the dynamic model of organizational knowledge creation process, the concept and abilities of wise leadership and phronesis (practical wisdom), and historical imagination and idealistic pragmatism. Professor Nonaka has won wide-ranging recognition for his work in developing the theory of Knowledge-based Management. In 2002, Professor Nonaka was conferred with a Purple Ribbon Medal by the Japanese government, and elected a member of the Fellows Group of the Academy of Management in the United States, becoming the first Asian scholar among the Group's members. Professor Nonaka was ranked number 20 in the Wall Street Journal's “Most Influential Business Thinkers (May 5, 2008).” In Autumn (2010), he was conferred with the Zuihōshō, or The Order of the Sacred Treasure, Gold Rays with Neck Ribbon, for outstanding achievement, and long service and contributions to education. In June 2012, Professor Nonaka received the Eminent Scholar Award from the Academy of International Business (AIB). In November 2013, he was presented with the Lifetime Achievement Award by Thinkers50, which is given to someone who has had a long-term impact on the way people think about and practice management. He was elected as a member of the Japan Academy in January 2016. Professor Nonaka received the Lifetime Achievement Award from the Haas School of Business at the University of California, Berkeley, on November 3, 2017, during the school's annual gala. Professor Nonaka has published many books and contributed numerous articles to management journals as well as other media both in Japanese and in English. Selected publications include: The Essence of Nation Management, Nihon Keizai Shimbun Shuppan-sha, 2014 (with co-authors); The Essence of Great Judgements, Diamond-sha, 2014 (with S. Ogino); The Grammar of Knowledge Creating Management for Prudent Capitalism, Toyokeizaishimpo-sha, 2012 (with N. Konno); Managing Flow: A Process Theory of the Knowledge-based Firm, Palgrave Macmillan, 2008 (with co-authors); The Essence of Strategy, Nihon Keizai Shimbun-sha, 2005 (with co-authors); Hitotsubashi on Knowledge Management, John Wiley & Sons, 2004 (with co-authors); The Essence of Innovation, Nikkei BP-sha, 2004 (with co-authors); Enabling Knowledge Creation: How to Unlock the Mystery of Tacit Knowledge and Release the Power of Innovation, Oxford University Press, 2000 (with co-authors); The Knowledge-Creating Company, Oxford University Press, 1995 (with H.Takeuchi); Strategic vs. Evolutionary Management: A U.S.-Japan Comparison of Strategy and Organization, Amsterdam: North-Holland, 1985 (with co-authors); “The Wise Leader” Harvard Business Review, May 2011 (with Hirotaka Takeuchi); “The Theory of the Knowledge-creating Firm: Subjectivity, Objectivity and Synthesis,” Industrial and Corporate Change, 14(3) 2005 (with R.Toyama); “Toward Middle Up-down Management: Accelerating Information Creation,” Sloan Management Review, Spring 1998; “Creating Organizational Order out of Chaos: Self-renewal in Japanese Firms,” California Management Review, Spring 1998; and “The Concept of ‘Ba’: Building a Foundation for Knowledge Creation,” California Management Review, 40 (3) 1998 (with N. Konno). The Knowledge-Creating Company, and Enabling Knowledge Creation each received a best book of the year award in business and management from the (Association of American Publishers, Inc) in 1996 and 2000, respectively. TAKEUCHI, Hirotaka: Professor of Management Practice, Harvard Business School Hirotaka Takeuchi currently teaches courses in the MBA and Executive Education programs. He is also Chair, Board of Trustees of International Christian University and Professor Emeritus of Hitotsubashi University. Professor Takeuchi received a BA from International Christian University in Tokyo, Japan, and an MBA and PhD from the University of California, Berkeley. Professor Takeuchi's first faculty position at Harvard was in the Marketing Unit from 1976 to 1983. Starting in 1983, he taught at Hitotsubashi University in Tokyo and served as the Founding Dean of its business school from 1998 to 2010. He returned to Harvard Business School in 2010 and serves as the Faculty Chair for Japan. Prior to his academic career, he worked at McCann-Erickson in Tokyo and San Francisco and at McKinsey & Company in Tokyo. Professor Takeuchi's research has focused on the knowledge creation process within organizations, the competitiveness of Japanese firms in global industries, and the link between strategy and innovation. He is the author or editor of 16 books, including The Knowledge-Creating Company: How Japanese Companies Create the Dynamics of Innovation co-authored with Ikujiro Nonaka (which won the 1995 Best New Book of the Year Award in the business and management category from the Association of American Publishers), Can Japan Compete? co-authored with Michael Porter, and Extreme Toyota: Radical Contradictions That Drive Success at the World's Best Manufacturer co-authored with Hitotsubashi professors Emi Osono and Norihiko Shimizu (which won the Best 30 Business Books by Soundview Executive Book Summaries in 2008). His recent Harvard Business Review articles are The Wise Leader (May 2011) and Embracing Agile (May 2016). Professor Takeuchi has received recognition in several business magazines over the years. Business Week voted him as one of the top 10 ″management school professors in demand for in-house corporate education programs” in the world (1993) and recognized him as one of “The Stars of Asia: 50 Leaders at the Forefront of Change” (2001). Fortune introduced him as “among the intellectual leaders of the younger, globally-minded generation that is coming to power in Japan (1996)." Fast Company selected him as one of the “Most Creative People in Business” (2016). Professor Takeuchi is or has been a member of the board of directors of Mitsui & Co, Daiwa Securities, ORIX, Integral, and three start-up companies, all based in Japan. He is also a director/trustee of several non-profit organizations, including Japan Society of Boston, Nonaka Institute of Knowledge, Ark Hills Club, International Christian University, and HLAB. He is or has been an adviser to Fast Retailing, All Nippon Airways, NTT DoCoMo, World Economic Forum, Japan Association of Corporate Directors, Japan Football Association, among others. He has been a member of a number of committees and councils formed by government agencies in Japan, including the Cabinet Office; Ministry of Economy, Trade and Industry; Ministry of Finance; Ministry of Education, Culture, Sports, Science and Technology; and Ministry of Land, Infrastructure and Transport and also a member of the editorial board of Japan Marketing Journal, Journal of Knowledge Management, and Hitotsubashi Business Review. 1 Knowledge is created through two interactive processes in SECI. For one, two types of knowledge – tacit and explicit knowledge – interact with each other in the epistemological dimension. For another, people who create knowledge interact with each other – within a team, within an organization, and inter-organizationally – in the ontological dimension.While we had the two dimensions in mind, in reality, we didn't incorporate the ontological dimension in the original SECI Model (Nonaka and Takeuchi, 1995). Consequently, we described the SECI process only in terms of the epistemological dimension as a conversion from tacit and explicit and vice versa: 1) Socialization: from tacit to tacit. 2) Externalization: from tacit to explicit. 3) Combination: from explicit to explicit. 4) Internalization: from explicit to tacit. 2 In 1983, Roger Schank, an educator, wrote an influential book Dynamic Memory, which focused on artificial intelligence. In 1999, he wrote Dynamic Memory Revisited, which focused on human intelligence, and reached a similar conclusion as ours – education should not be about explicit knowing but about doing things. 3 As a side note, the tentative working title of our sequel book was actually “The Knowledge-Practicing Company,” but we did not follow through with it, thinking it reflected linear thinking. We concluded that replacing Creating with Practicing was too logical and deductive for this messy world we live in today, which is why we decided to make an inductive jump by replacing the centerpiece of our title, namely Knowledge with Wisdom. 4 The multi-dimensionality of the SECI Spiral Model depicted in Fig. 1 works as follows. The base, shown on the horizontal axis, is made up of tacit and explicit knowledge, which represents the two ends of the epistemological dimensions. The ontological dimension is shown on the vertical axis, where each SECI cycle moves up to a higher-level ontological base, from the individual level to the organization, to the inter-organizational, the community, and finally, societal levels. The model shows that the knowledge base becomes enlarged when SECI goes through one horizontal cycle, as well as when SECI spirals up to higher ontological levels over time. 5 As shown in Fig. 2 (Nonaka and Takeuchi, 2019: 62), individual knowledge is socialized and externalized into team knowledge; team knowledge is combined into organizational knowledge; and organizational knowledge is internalized back into (upgraded) individual knowledge, while interacting with the environment throughout the conversion process. 6 According to Teece, dynamic capabilities refers to the particular (non-imitable) capacity business enterprises possess to shape, reshape, configure, and reconfigure assets so as to respond to changing technologies and markets and escape the zero profit condition. Dynamic capabilities relate to the enterprise's ability to sense, seize and adapt, in order to generate and exploit internal and external enterprise-specific competences and to address the enterprise's changing environment. 7 Keith Yamashita, “Neuroplasticity: We Can Change Our Brain,” an unpublished document prepared for Global Treehouse by SY Partners. Yamashita is the founder and chairman of SY Partners, which has more than a quarter of a century history of helping leaders and organizations transform with its base in San Francisco and New York. 8 Joichi Ito, “Resisting Reduction: A Manifesto,” in Joichi Ito (ed.) Resisting Reduction: Designing Our Complex Future with Machines (Cambridge, MA: MIT Press, pending) p. 11. 9 Steve Denning, “A Roadmap for Reshaping Capitalism,” an unpublished report of The New Enlightenment Conference, held on July 1–2, 2019 at Penmure House, Edinburgh Scotland. p. 8. 10 As a matter of fact, in 2019 the World Economic Forum suggested to stop using the word “capitalism” and replacing it with “talentism.” 11 We are grateful of Professor Robert M. Grant of Bocconi University, Department of Management and Technology, for his helpful comments in reviewing this paper. ==== Refs References Brandt A. Eagleman D. The Runaway Species: How Human Creativity Remakes the World 2017 Catapult New York Freedman L. Strategy: A History 2013 Oxford University Press New York Funabashi H. The Wisdom of Our Ancestors 2013 The Japan Journal Tokyo Herman A. Skiing Fortune 15 2015 March Kerr J. Legacy: what the All Blacks Can Teach Us about the Business of Life 2013 Constable London Koch C. Consciousness: Confessions of a Romantic Reductionist 2012 MIT Press Cambridge, MA Morson G.S. Schapiro M. Cents and Sensibility 2017 Princeton University Press Princeton, NJ Nonaka I. Takeuchi H. The Knowledge-Creating Company: How Japanese Companies Create the Dynamics of Innovation 1995 Oxford University Press New York Nonaka I. Takeuchi H. The Wise Company: How Companies Create Continuous Innovation 2019 Oxford University Press New York Osono E. Shimizu N. Takeuchi H. Extreme Toyota: Radical Contradictions that Drive Success at the World's Best Manufacturer 2008 John Wiley & Sons Hoboken, NJ Polanyi M. The Tacit Dimension 1996 Doubleday Garden City, NY Porter M. Competitive Strategy: Techniques for Analyzing Industries and Competitors 1980 Free Press New York Porter M. Takeuchi H. Can Japan Compete? 2000 Basic Books New York Rother M. Toyota Kata: Managing People for Improvement, Adaptiveness, and Superior Results 2010 McGraw Hill New York Safian R. Fast Company November 2014 Find Your Mission Sax D. The Revenge of Analog: Real Things and Why They Matter 2016 Public Affairs New York Teece D. Dynamic Capabilities and Strategic Management: Organizing for Innovation and Growth 2009 Oxford University Press New York Varela F. Thompson E. Rosch E. 1st MIT Press pbk The Embodied Mind: Cognitive Science and Human Experience 1993 MIT Press Cambridge, MA

0

PMC9753931

NO-CC CODE

2022-12-16 23:26:22

no

Long Range Plann. 2021 Aug 8; 54(4):102070

utf-8

Long Range Plann

10.1016/j.lrp.2021.102070

oa_other

==== Front Lancet Lancet Lancet (London, England) 0140-6736 1474-547X Elsevier Ltd. S0140-6736(22)00117-9 10.1016/S0140-6736(22)00117-9 Perspectives Pregnancy and the origins of illness Lyerly Anne Drapkin a a Department of Social Medicine, University of North Carolina at Chapel Hill, Chapel Hill, NC 27599-7240, USA 28 1 2022 29 January-4 February 2022 28 1 2022 399 10323 428429 Richardson Sarah S The Maternal Imprint: The Contested Science of Maternal-Fetal Effects2021University of Chicago Press9780226544809 376US$26·00© 2022 Elsevier Ltd. All rights reserved. 2022 Elsevier Ltd Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. ==== Body pmcThe effects of the COVID-19 pandemic are expected to be felt for years after the virus is contained. Alongside the more than 5·5 million COVID-19 deaths globally, there have been massive shifts in how we understand and practise medicine and public health, transformations of global economies, and losses in livelihoods, health, education, and security. There is also the disquieting prospect that our current trauma could have lasting intergenerational impacts—that children gestated or born during this pandemic might carry an imprint of their mother's experience with lifelong and inheritable consequences. It is this notion—the “bewitching idea that the environment in which you are gestated leaves a permanent imprint on you and your future generations”—that Sarah Richardson explores in her enlightening book, The Maternal Imprint: The Contested Science of Maternal-Fetal Effects. Richardson, a renowned historian and philosopher of science, begins her book with research on the intergenerational impacts of the Holocaust. Citing psychiatrist and neuroscientist Rachel Yehuda's studies of Holocaust survivors and their children, Richardson introduces the burgeoning field of developmental origins of health and disease (DOHaD) and its use of epigenetics—analyses of heritable changes that affect gene expression but not DNA sequence—as a means to measure and characterise the impact of the gestational environment. In recent years, epigenetic changes traced to gestation have been linked, if tenuously, to a child's future risk of conditions ranging from obesity to asthma to poor response to stress. In a study now widely cited in academic and social contexts, Yehuda measured levels of DNA methylation at a particular genetic locus in Holocaust survivors and their adult children, and found them to diverge in opposite directions when compared with controls. She and her team pointed to such differences as “support [for] an intergenerational epigenetic priming of the physiological response to stress in offspring of highly traumatized individuals”. The data have been taken up as evidence that trauma can be genetically encoded and inherited, thereby “situating”, as Richardson puts it, “the maternal-fetal interface at the nexus of inter-generational trauma”. How should we think about the rise of a new science that implicates maternal bodies in this way, especially in the present context of COVID-19? We know that SARS-CoV-2 infection in pregnancy is associated with increased risks of severe COVID-19 and adverse pregnancy outcomes. Despite these risks, some pregnant individuals have not taken up COVID-19 vaccines, partly due to concerns about vaccines untested in pregnant populations. Infection aside, no-one is immune to the wide-ranging stressors that the pandemic has entailed. Although vertical transmission of SARS-CoV-2 is very rare, to what extent might maternal bodies be understood as vectors in the transmission of our collective loss? Although COVID-19 is not mentioned in The Maternal Imprint, Richardson speaks with clarity to such questions, and to policy makers, scientists, physicians, and pregnant people and parents who will inevitably encounter them. Deploying the tools of history, philosophy, and gender studies of science, Richardson's elegant analysis of the long history and current status of efforts to study so-called maternal effects reveals scientific claims about the “long reach of the womb [to be] at once beguiling, challenging to validate, stubbornly persistent once launched, and beset by scientific controversy” and suggests we ought to interpret such claims with a fair amount of caution. First is the problem of crypticity— a term Richardson helpfully uses to describe the tenuousness of the links between cause and effect that characterise maternal effects science. Unlike teratology, in which the impact of a prenatal exposure can be immediately observable, as exemplified by major congenital defects linked to the drug thalidomide, the effects of interest in DOHaD studies are typically small, influenced by ongoing social processes and environments (eg, obesity), difficult to measure (eg, response to stress), and often manifest at a temporal distance from the initial exposure. As Richardson's rich history of maternal effects science makes clear, these are not new challenges—they are emblematised, for instance, by efforts in the 1960s and 1970s to link birthweight, race, and social inequality, to which she attends in detail. Moreover, the challenges of crypticity are not redressed by epigenetic technologies—even, as she argues in a robust chapter-long critique of the emerging science, among the boldest new research programmes. What is particularly compelling about Richardson's approach is that by situating the field's challenges in a broader social and historical context, we get a sense of why they matter. Crypticity matters, she argues, because it requires a “permissive approach” to constructing narratives of causation, and gives latitude to scientists to ask provocative questions about the impact of developmental exposures on the health of generations to come. Perhaps more importantly, crypticity creates space for social assumptions and values to enter the chain of causal reasoning, which can be powerfully distorting where pregnant people are concerned. A second reason for caution, then, is that in maternal effects science, social assumptions can have an outsize and distorting role because they concern the gendered politics of reproduction. Given the entanglement of reproduction with broad political, economic, and cultural institutions and ideologies, such assumptions can reflect and amplify inequality and oppression when applied to scientific questions about the maternal–fetal interface. One of the most powerful social assumptions is that mothers are responsible for the outcomes of pregnancy. In many contexts, the tendency to ascribe responsibility to mothers has led to reams of everyday advice directed at pregnant people, which may lack robust evidence or accentuate precaution, as well as blaming mothers for pregnancy outcomes well beyond their control. This stance has also led to a range of ethical violations, including the incarceration of women for adverse pregnancy outcomes and efforts to restrict the autonomy of pregnant individuals who do not follow medical advice or appear not to act in the best interests of their offspring. Maternal effects science manifests at least two additional concerns. The first is that a focus on maternal responsibility allows for the accumulation of causal attributions in DOHaD principally around pregnancy and maternal behaviours, but with insufficient data to establish causation. The second concern is that these assumptions contribute to a “drastically limited”, according to Richardson, picture of influences on development since much DOHaD research does not adequately consider paternal, postnatal, and other social and environmental factors that may influence the long-term health of offspring. For all the attention on maternal responsibility, a third reason for caution is that the language and emphasis of maternal effects science undermines the view of mothers as persons. It is a problem to which Richardson speaks in the epilogue, reflecting on contemporary analyses of epigenetics in which, “the environment is represented by the fuzzy, receding figure of the maternal”. Over time, as Richardson relates, pregnant bodies have been described as “sensitive biological barometers of the social environment”, as “transducer[s]…[at] risk of malfunction”, as “epigenetic vectors”, and, frequently still, as the “gestational…[or] developmental environment”. More than a semantic offence, characterisations of maternal bodies as environments have given rise to substantive concerns of real consequence to women, such as failures to appropriately weight maternal health in recommendations for public policies and clinical care or consider the interests of pregnant people in public health research. As such, the use of the term “gestational environment” in the context of DOHaD prompts questions about whether and how the interests of mothers themselves fit into the research agenda. Ultimately, Richardson sounds a much-needed alarm about the potential for DOHaD science to “function as another stigmatizing, determinist discourse that threatens women's reproductive autonomy and contributes to stigma and moral panic about the mothering practices of poor women, women of color, and non-normative mothers of all kinds”. These cautions aside, I remain beguiled by fetal origins science. For despite the myopic focus on the womb that has largely characterised its uptake, DOHaD research offers an important, if yet unrealised, corrective. As Richardson notes, the proximate foundations of fetal origins science can be traced to the work of British epidemiologist David Barker who postulated that conditions of poverty during gestation can lead to heart disease in adulthood. Indeed, contexts of deprivation or disaster—the Holocaust, the Dutch famine, a major ice storm in Quebec—continue to ground the field's most prominent research streams. To the extent that DOHaD research has consistently looked outside the maternal body for sources of harm, it reminds us of the importance of the social determinants of health, and that the “gestational environment” is best understood not as the womb but as the much broader environment in which a person gestates. And it is that broader environment—rather than maternal behaviours or choices—towards which we might best direct ameliorative efforts, such as by mitigating the harms of climate change, racism, and a range of structural inequities and inequalities. These efforts are especially relevant in the context of the collective trauma of and the immense disparities laid bare during the COVID-19 pandemic. In the likelihood that epigenetics might be used to explore the long shadow of our current trauma, Richardson's book offers an important lens through which we can regard claims about the role of maternal bodies, highlighting reasons for caution. As Richardson notes, some regard epigenetics as evidence of an “embodied mechanism of memory”—a way for the body to convey and perhaps honour the past. She reminds us too that fetal programming science must also recognise and honour the interests and experiences of those who may or may not, through gestation, pass such histories along. ==== Refs Further reading Erikson KT Everything in its path: destruction of community in the Buffalo Creek flood 1976 Simon & Schuster New York Iacobucci G Covid-19 and pregnancy: vaccine hesitancy and how to overcome it BMJ 375 2021 n2862 Jamieson DJ Rasmussen SA An update on COVID-19 and pregnancy Am J Obstet Gynecol 2021 published online Sept 14. 10.1016/j.ajog.2021.08.054 Lyerly AD Mitchell LM Armstrong EM Risk and the pregnant body Hastings Cent Rep 39 2009 34 42 20050369 Lyerly AD Little MO Faden RR A critique of the “fetus as patient” Am J Bioeth 8 2008 42 44 Rosner E Survivor cafe: transforming the legacy of intergenerational trauma 2017 Counterpoint Berkeley, CA Paltrow LM Flavin J Arrests of and forced interventions on pregnant women in the United States, 1973–2005: implications for women's legal status and public health J Health Polit Policy Law 38 2013 299 343 23262772 Rivkin-Fish M Buchbinder M Walker R Understanding health inequalities and justice 2016 University of North Carolina Press Chapel Hill, NC Silver RC Holman EA Garfin DR Coping with cascading collective traumas in the United States Nat Hum Behav 5 2021 4 6 33106630 Wei SQ Bilodeau-Bertrand M Liu S Auger N The impact of COVID-19 on pregnancy outcomes: a systematic review and meta-analysis CMAJ 193 2021 E540 E548 33741725 WHO Definition and categorization of the timing of mother-to-child transmission of SARS-CoV-2: scientific brief, 8 2021 World Health Organization Geneva Yehuda R Daskalakis NP Bierer LM Holocaust exposure induced intergenerational effects on FKBP5 methylation Biol Psychiatry 80 2016 372 380 26410355 Yuko E COVID-19 is traumatizing all of us. How will we cope after it's over? Rolling Stone May 5, 2020 Richardson SS Daniels CR Gillman MW Society: don't blame the mothers Nature 512 2014 131 132 25119222

0

PMC9753932

NO-CC CODE

2022-12-16 23:26:22

no

Lancet. 2022 Jan 28 29 January-4 February; 399(10323):428-429

utf-8

Lancet

10.1016/S0140-6736(22)00117-9

oa_other

==== Front Surgery Surgery Surgery 0039-6060 1532-7361 Elsevier Inc. S0039-6060(21)00152-5 10.1016/j.surg.2021.02.032 Letter to the Editor Re: “Colorectal screening: We have not caught up. A surge of colorectal cancer after the coronavirus disease 2019 (COVID-19) pandemic?” Mak Victoria P. Department of Quantitative Health Sciences, John A. Burns School of Medicine and University of Hawaii Cancer Center, University of Hawaii at Mānoa, Honolulu, Hawaii Le Marchand Loïc University of Hawaii Cancer Center, University of Hawaii at Mānoa, Honolulu, Hawaii 27 3 2021 7 2021 27 3 2021 170 1 348349 © 2021 Elsevier Inc. All rights reserved. 2021 Elsevier Inc. Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. ==== Body pmcTo the Editors: We read with interest the article by Challine et al (2021)1 comparing the number of colonoscopies during the coronavirus disease 2019 (COVID-19) lockdown and postlockdown periods in France with those from the previous 2 years. The comparison clearly highlighted the decrease in colorectal cancer (CRC) screenings during the lockdown, without any compensatory postlockdown increase, predicting a rise in undiagnosed colorectal cancers. Data were collected from a national database with mandatory reporting from both public and private hospitals, allowing for a comprehensive analysis of health data in France. However, the article did not categorize the number of colonoscopies by race/ethnicity or socioeconomic status, which would allow for a detailed comparison with findings from other countries. In the United States, African Americans, Native Americans, and other underprivileged minorities, often with less access to quality healthcare, preemptive screenings, and a healthy diet, suffer from lower survival for all stages of CRC.2 Cancer screening is likely to play a major role in this disparity since, from 1975 to 2015, the incidence of CRC in the United States decreased by 21% among African Americans (from 56.9 to 44.7 per 100,000), compared to a 40% decrease in Whites (60.2 to 36.2).2 , 3 Additionally, the association between social measures (socioeconomic status, race/ethnicity) and CRC mortality was examined comparing behavioral and medical preventive factors over time with population-based CRC mortality trend data in the United States.4 A lower socioeconomic status, as well as race/ethnicities of African American, Hispanic, Asian/Pacific Islander, and Native American, were found to be associated with decreased access to age-appropriate CRC screening.4 The decrease in colonoscopies described by Challine et al during the current COVID-19 pandemic foreshadows a rise in CRC mortality. Future studies of delay in screening practices should collect data on ethnic/racial minorities and low socioeconomic status groups to identify disparities and provide the information needed to equitably address this important public health challenge through targeted interventions. Funding/Support VPM is funded by US National Institutes of Health grant R25 CA244073. Conflict of interest/Disclosure The authors have no disclosure. ==== Refs References 1 Challine A. Lazzati A. Dousset B. Voron T. Parc Y. Lefevre J.H. Colorectal screening: we have not caught up. A surge of colorectal cancer after the coronavirus disease 2019 (COVID-19) pandemic? Surgery 2021 S0039-6060(20)30856-4 2 Rawla P. Sunkara T. Barsouk A. Epidemiology of colorectal cancer: incidence, mortality, survival, and risk factors Prz Gastroenterol 14 2019 89 103 31616522 3 SEER∗Explorer: An interactive website for SEER cancer statistics [Internet]. Surveillance Research Program 2017 National Cancer Institute 4 Clouston S.A.P. Acker J. Rubin M.S. Chae D.H. Link B.G. Fundamental social causes of inequalities in colorectal cancer mortality: a study of behavioral and medical mechanisms Heliyon 6 2020 e03484

33781585

PMC9753933

NO-CC CODE

2022-12-16 23:26:22

no

Surgery. 2021 Jul 27; 170(1):348-349

utf-8

Surgery

10.1016/j.surg.2021.02.032

oa_other

==== Front Surgery Surgery Surgery 0039-6060 1532-7361 Elsevier Inc. S0039-6060(21)00213-0 10.1016/j.surg.2021.03.021 Letter to the Editor Colorectal screening: We have not caught up. A surge of colorectal cancer after the coronavirus disease 2019 (COVID-19) pandemic? Challine Alexandre MD Department of Digestive and General Surgery, Sorbonne University, Saint-Antoine Hospital, Paris, France INSERM, UMR_S 1138, Université Paris Descartes, Centre de Recherche des Cordeliers, France Lazzati Andrea MD, PhD Department of General Surgery, Centre Hospitalier Intercommunal de Créteil, France University Paris Est-UPEC, Créteil, France Katsahian Sandrine MD, PhD Epidemiology Department, University of Paris, European Hospital Georges Pompidou, France INSERM, UMR_S 1138, Université Paris Descartes, Centre de Recherche des Cordeliers, France Parc Yann MD, PhD Lefevre Jeremie H. MD, PhD ∗ Department of Digestive and General Surgery, Sorbonne University, Saint-Antoine Hospital, Paris, France ∗ Corresponding author. 21 4 2021 7 2021 21 4 2021 170 1 349350 6 3 2021 © 2021 Elsevier Inc. All rights reserved. 2021 Elsevier Inc. Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. ==== Body pmcTo the Editor: We read with interest the letter from Mak et al1 concerning our study on the drop in colonoscopies during the lockdown in France.2 Indeed, we did not perform analyses concerning ethnicity or race, as such data are unavailable in France for legal and ethical reasons. However, to gain insight into the role of social status, we performed a second analysis. Our national database (Programme de médicalisation des systèmes d'information), used in the article, provides the zip codes of patients’ residential communes. These codes are associated with the French deprivation (Fdep) index, which reports the level of precarity of each zip code.3 , 4 The calculation of this index is based on the unemployment rate, the worker rate, schooling rate, and monthly income. The higher the index the more disadvantaged is the zip code and, therefore, the commune. During the lockdown period, the Fdep index increased significantly for patients who underwent colonoscopy compared to the prelockdown period (lockdown: 0.21 [95% confidence interval {CI}, –0.66; 0.90] vs before lockdown: 0.10 [95% CI, –0.81; 0.84]; P < .001). This reflects the fact that the ratio of colonoscopies in socioeconomically deprived communes was higher than in wealthier zones during the lockdown. For the most deprived communes, the ratio of performed colonoscopies increased (before lockdown: 25% vs lockdown 27%). Comparatively, the ratio of performed colonoscopies in the most wealthier communes decreased (before lockdown: 25% vs lockdown 22%). Deprived communes were not the most impacted by the decrease of activity. Moreover, the Fdep was not significantly different before and after the lockdown (before lockdown: 0.10 [95% CI, –0.81; 0.84] vs after lockdown: 0.10 [95% CI, –0.80 to 0.83]; P = .66). The first explanation of this finding is that the major decrease of activity was observed in the largest communes, with a low Fdep, where university hospitals and large private clinics stand. Interestingly, there was no catch-up during the postlockdown period for nondeprived communes, as the index did not decrease significantly. A second explanation is the characteristics of the French health care system, which provides almost universal medical coverage for the entire population, reducing the impact of socioeconomic criteria on the screening or treatment of medical diseases. We agree with the comment regarding a possible increased mortality of colorectal cancer due to more advanced stages. In a previous study, we did not report association between digestive elective surgery during the lockdown and postoperative death.5 Only a coronavirus disease 2019 infection was a risk factor for postoperative mortality. Further analyses are needed to evaluate the impact of the different lockdowns and the global coronavirus disease 2019 pandemic on colorectal cancer treatment with a long-term follow-up. The rate of patients requiring adjuvant chemotherapy after resection may be a good surrogate marker to be evaluated. Funding/Support No funding to report. Conflict of interest/Disclosure No conflicts of interest to report. ==== Refs References 1 Mak V.P. Le Marchand L. RE: “Colorectal screening: We have not caught up. A surge of colorectal cancer after the coronavirus disease 2019 (COVID-19) pandemic?” Surgery 170 2021 348 349 33781585 2 Challine A. Lazzati A. Dousset B. Voron T. Parc Y. Lefevre J.H. Colorectal screening: We have not caught up. A surge of colorectal cancer after the coronavirus disease 2019 (COVID-19) pandemic? Surgery 169 2021 991 993 33485642 3 Rey G. Jougla E. Fouillet A. Hemon D. Ecological association between a deprivation index and mortality in France over the period 1997-2001: variations with spatial scale, degree of urbanicity, age, gender and cause of death BMC Public Health 9 2009 33 19161613 4 Windenberger F. Rican S. Jougla E. Rey G. Spatiotemporal association between deprivation and mortality: trends in France during the nineties Eur J Public Health 22 2012 347 353 21459841 5 Challine A. Dousset B. de’Angelis N. Impact of coronavirus disease 2019 (COVID-19) lockdown on in-hospital mortality and surgical activity in elective digestive resections: A nationwide cohort analysis Surgery 2021 10.1016/j.surg.2020.12.036 Available from:

33892951

PMC9753934

NO-CC CODE

2022-12-16 23:26:22

no

Surgery. 2021 Jul 21; 170(1):349-350

utf-8

Surgery

10.1016/j.surg.2021.03.021

oa_other

==== Front Appl Geogr Appl Geogr Applied Geography (Sevenoaks, England) 0143-6228 0143-6228 Elsevier Ltd. S0143-6228(21)00034-5 10.1016/j.apgeog.2021.102418 102418 Article Impact of the COVID-19 lockdown on air quality in the Delhi Metropolitan Region Roy Shouraseni Sen a∗ Balling Robert C. Jr. b a Department of Geography and Regional Studies, University of Miami, FL, USA b School of Geographical Sciences and Urban Planning, Arizona State University, AZ, USA ∗ Corresponding author. 1 3 2021 3 2021 1 3 2021 128 102418102418 10 6 2020 4 1 2021 27 1 2021 © 2021 Elsevier Ltd. All rights reserved. 2021 Elsevier Ltd Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. With the rapid spread of COVID-19 related cases globally, national governments took different lockdown approaches to limit the spread of the virus. Among them, the Government of India imposed a complete nationwide lockdown starting on March 25, 2020. This presented a unique opportunity to explore how a complete standstill in regular daily activities might impact the local environment. In this study, we have analyzed the change in the air quality levels stemming from the reduced anthropogenic activities in one of the most polluted cities in the world, the Delhi Metropolitan Region (DMR). We analyzed station-level changes in particulate matter, PM10 and PM2.5, across the DMR between April 2019 and 2020. The results of our study showed widespread reduction in the levels of both pollutants, with substantial spatial variations. The largest decreases in particulate matter were associated with industrial and commercial areas. Highest levels of PM10 and PM2.5 were observed near sunrise with little change in the time of maximum between 2019 and 2020. The results of our study highlight the role of anthropogenic activities on the air quality at the local level. Keywords COVID-19 PM10 PM2.5 Delhi metropolitan region Harmonics Diurnal ==== Body pmc1 Introduction Within a few months in 2020, COVID-19 had massively impacted the entire globe far beyond the direct effects on human health. These impacts vary at different spatial scales as reflected in the number of infections and mortality, as well as the indirect impacts on the economy and environment. Many local and national governments implemented various degrees of lockdown or “shelter at home” policies to contain the spread of this global pandemic. This has resulted in a massive reduction in global economic activity thereby significantly lowering air pollution levels in many areas of the world including China (Wu et al., 2020) and India (Patel et al., 2020). Specifically, substantial declines in ground level nitrogen dioxide, ozone, and particulate matter were reported within the first two weeks of the lockdown in 27 countries as revealed from the analysis of satellite data during February and March 2020 (Venter et al., 2020). However, due to the coarse spatial and temporal resolution of the satellite data, their study was limited in revealing the local level patterns in air pollution. This is particularly relevant in parts of Asia, such as in India, where most of the high pollution levels are concentrated in the large urban areas. Therefore, in the present study we have analyzed the change in levels of two pollutants, PM10 and PM2.5, over a one month period between April 2019 and 2020 in the Delhi Metropolitan Region (DMR), India (Fig. 1 ).Fig. 1 Distribution of air pollution monitoring stations in the Delhi Metropolitan Region (DMR). Fig. 1 The Government of India imposed lockdown on all of its 1.3 billion citizens on March 25, 2020 to limit the spread of the COVID-19; the lockdown ended after 55 days on May 19, 2020. As a result, there was widespread shutdown of factories, road, and air traffic across the DMR. The shutdown was so severe that it triggered a massive migration of wage laborers walking for several 100s of kilometers across northern India to their villages caused by no available jobs and the cancellation of long distance transportation such as railways and bus systems. The initial results from satellite image analysis of aerosol optical depths revealed a significant drop across the northern plains resulting from the closure of coal-fired heavy industries and reduction in traffic across large urban areas like the DMR (Patel et al., 2020). In addition, there have been widespread reports in mainstream media outlets about the reduction in pollution levels in the DMR dropping from unhealthy and hazardous to good (Business Today, 2020; Ellis-Peterson et al., 2020). These trends have also been documented in recently published studies. For instance, Shehzad et al. (2020) analyzed Sentinel – 5 P satellite images and reported a significant improvement in air quality and a 40–50% decrease in atmospheric nitrogen dioxide levels in the large cities of India, including DMR and Mumbai. More detailed analysis of station level data for the DMR during pre and post lockdown periods revealed significant decreases in levels of particulate matter within days of the lockdown (Kumari & Toshniwal, 2020; Mahato et al., 2020). Additionally, detailed hourly analysis of air pollutants revealed a decline in the levels of particulate matter during nighttime and peak traffic hours (Singh et al., 2020). This is particularly significant in view of the DMR and its suburbs being ranked as the most polluted city in the world for at least the last five years. Specifically, the pollutants with consistently high levels in the DMR are PM10, PM2.5, NO2, SO2, and O3 (Aneja et al., 2001; Balachandran et al., 2000; Gurjar et al., 2004; Kandikar 2007; Nagar et al., 2019; O'Shea et al., 2015; Sahu et al., 2011). Furthermore, India has recorded an increasing trend in population-weighted mean concentrations of PM2.5, with a noticeable increase since 2010 (Bhakta et al., 2019; Gurjar et al., 2016). These elevated concentrations of PM2.5 exposures have been attributed to annual premature death estimates of 272,000 for chronic obstructive pulmonary disease (COPD), 110,600 for ischemic heart disease, and 14,800 for lung cancer (Chowdhury & Dey, 2016). Specifically, for the DMR a reduction in life expectancy due to exposure to PM2.5 was 6.3 ± 2.2 years greater than the same for overall Indian population (3.4 ± 1.1 years) (Ghude et al., 2016). Due to the paucity of data, most of the above mentioned station level studies in the DMR have examined the variations for select pollutants in limited number of stations, ranging between two and five. The results from these studies revealed significant variations in the levels of various pollutants and diurnal cycles across the study area. All of these studies highlighted the excessive readings and increase in the levels of pollutants in the DMR, mainly resulting from traffic congestion and industrial activities. Thus, with the implementation of the strict measures associated with the lockdown, the drop in air pollution levels is distinctly visible and merits detailed analysis. Therefore, the two specific objectives of the study are:1. Determine the local-level spatial patterns of change in the levels of two major pollutants, PM10 and PM2.5, between April 2019 and 2020. 2. Analyze the changes in the time of maximum for peak levels in PM10 and PM2.5 between April 2019 and 2020. 2 Data and methods Station-level hourly data for PM10 and PM2.5 were obtained from the Central Pollution Control Board (CPCB), collected as part of the Air Quality Monitoring Program in India. The data were collected for April 2019 and 2020 to assess the change in pollution levels due to the lockdown. There are 40 pollution monitoring stations spread across the DMR, out of which there were continuous data available for 34 stations for PM10 and 31 stations for PM2.5 (Fig. 1). We analyzed only two pollutants PM10 and PM2.5 because these were the only two pollutants with complete data across all the stations. The data for the other pollutants were incomplete and thus not suitable for analysis. Furthermore, these two pollutants have consistently been the biggest challenge for air quality scientists and policymakers and therefore the results of the study are particularly relevant. In order to test the spatial distribution of the station network across the study area, the nearest-neighbor statistics were calculated as the ratio between the observed mean distance among the station locations and the expected mean distance given a random distribution (Clark & Evans, 1954). The ratios of 1.03 for PM10 and 0.90 for PM2.5 show that both station networks have a random distribution. We analyzed the changes in the levels of PM10 and PM2.5 for the monthly average values by calculating the differences at the station level. The percentage change in the levels of the pollutants at the station level were visualized using spatial interpolation, simple kriging. Kriging is a surface interpolation method utilized to visualize spatial variation through a variogram, thus minimizing prediction errors (Oliver & Webster, 1990; Sen Roy, 2006a). The final interpolated surface is calculated by incorporating the spatial and statistical relationships among the different variables by using the following equation:Z(s) = μ + ε(s) Where μ is a known constant utilized to interpolate the resulting surface, s denotes the location being predicted, and ε(s) is the error term (Sen Roy, 2006b). This method was preferred because of its accuracy in surface interpolation and lower root mean square error. Harmonic analysis was used to determine the time of maximum concentration of PM10 and PM2.5 levels based on the average hourly values over the course of a month. To fit a trigonometric wave with one maxima and one minima, the harmonic equation for any station with 24 values takes the form:f(x)=X‾+∑r=1N/2Arcos(rθ−Φr) where f(x) is the estimated value in each interval, X‾ is the average value over the N = 24 intervals, A r is the amplitude of the rth harmonic wave, r is the frequency or number of times the harmonic wave is repeated over the fundamental period (in this case r = 1), θ is derived as 2πx/N where x signifies the intervals over the fundamental period, and Φris the phase angle of the rth harmonic reinterpreted as the time of maximum. The basic form is explained below:f(x)=X‾+∑r=1N/2[arsin(2πrx/P)+brcos(2πrx/P)] where the Fourier coefficients, a r and b r, are calculated as:ar=∑x=1N2N[f(x)sin(2πrx/P)] andbr=∑x=1N2N[f(x)cos(2πrx/P)] The amplitude, A r, is calculated as (a r 2 + b r 2 ) 0.5, the phase angle, Φr, equals tan−1 (a r /b r), and the portion of variance denoted by the rth harmonic wave, V r, is calculated as A r 2 /2s 2 wherein s is the standard deviation of the N values (see Nelson, 1983, p. 190). Given that the harmonic wave is fitted to PM10 or PM2.5 averages in 24 hourly intervals, the explained variance levels had to be > 0.16 to be statistically significant at the 0.05 level of confidence. This method has been successfully used to study diurnal patterns in air pollution variables in previous studies (Liu & Sen Roy, 2014; O'Shea et al., 2015). 3 Results and discussion The DMR is located in the northern interior of the Indian subcontinent, and thus experiences a typical continental climate. The comparative results from the two different years reveal substantially lower levels of in the levels of air pollution between the two years with distinct spatial variations across the DMR (Fig. 2 ). Typical of megacities of Asia, the DMR is densely populated (greater than 25,000 persons per km2), with an estimated total population of about 17 million. Majority of the DMR is urban, with 97.5% of the population identified as urban (Sen Roy et al., 2020). According to Jain et al. (2016), the net percent change in land use from 1977 to 2014 for urban built-up areas increased by 30.61%, along with a decrease by 22.75% for cultivated areas, 5.31% for dense forest, and 2.76% for wasteland. With the steep increase in population accompanied by car ownership and unplanned urbanization, the DMR has experienced steep increase in levels of air pollution over the years. The extremely hazardous levels of air pollution in the winter months has led to the forced closure of schools and steep increase respiratory diseases among the local population. The sources of air pollution are not only from local sources such as vehicular emissions and industrial activities, but more recently the burning of crops after harvesting in the agricultural fields in the neighboring states of Punjab and Haryana. Thus, the pollution levels noted in 2019 typifies that time of the year in the DMR over the last decade.Fig. 2 Changes in average hourly levels of particulate matter between April 2019 and 2020 (a) PM10; (b) PM2.5. Fig. 2 As is evident from Fig. 2, the level for both pollutants in 2019 were higher than those observed in 2020 for all the stations. The percentage change for PM10 ranged between 20 and 70%, while the range of decline was wider for PM2.5 at 15–90%. At the local level, 21 out of 31 stations experienced greater than 50% decline for PM10. The largest declines were concentrated in the eastern half of the DMR, which also consistently experienced higher levels of particulate matter. Moreover, for the PM2.5 the largest percentage declines (greater than 50%) were observed for only 8 out of 34 stations located in the central and eastern parts of the DMR. The greater decline in central DMR is due to the lower levels of economic activity associated with the high density of office buildings and commercial areas in the core downtown area. Similarly, the higher rates of decline in the eastern DMR are associated with the closure of local factories and thermal power plants. Specifically, there are three thermal power plants located in the DMR, which include Indraprastha Power Station and Rajghat Power House in the east, and Badarpur Power Station in the northwest. Majority of the land use in the northern and western DMR is residential, and thus experienced relatively lower levels of decline in the pollutants. Two stations, located in Shadipur (west of central downtown) and Dilshad Garden (east) experienced greater than 85% decline in PM2.5 (Fig. 2b and Table 2 ). The lowest differences in average hourly levels of PM10 were located outside the central core of the DMR, such as Najafgarh (192 μg/m3 in 2019 vs 146.1ug/m3 in 2020), Ashok Vihar (168.2ug/m3 in 2019 vs 95.42μg/m3 in 2020), and Aya Nagar (155.7ug/m3 in 2019 vs 78.53μg/m3 in 2020) (Table 1, Fig. 2a). Among these three stations, two of them are predominantly residential, while Najafgarh consists of transitioning from rural to urban land uses mixed with industries. Similarly in the case of PM2.5, the lowest differences in average hourly levels were observed in predominantly residential areas, including Lodhi Road (57μg/m3 in 2019 vs 46.5ug/m3 in 2020), Aya Nagar (51.8ug/m3 in 2019 vs 39.1ug/m3 in 2020), and Punjabi Bagh (155.7ug/m3 in 2019 vs 73.4ug/m3 in 2020) (Table 2, Fig. 2b). Moreover, the hourly maximum values between the two years revealed higher values across most of the stations during 2019 for both PM10 and PM2.5, except in Punjabi Bagh (for both pollutants) and Pusa DPCC (PM2.5) in West Delhi (Table 1, Table 2). There was also a greater amount of variance in the hourly maximum levels for both of the pollutants during 2019 compared to 2020. However, the patterns were not as distinct in the case of hourly minimum values across the DMR. Our results are in conformity with the results of a previous study by Tiwari et al. (2015), who found substantial spatial variations in the correlation between observed levels of PM10 and PM2.5 at the seasonal scale and weekday vs weekends for a limited number of stations. Specifically, their analysis of hourly level observations of these two particulates revealed the mean coarse mode particulate mass concentration (PM10–2.5) as 113.6 ± 70.4 μg/m3 (varied from 13.6 to 630 μg/m3) constituting about 49% contribution of PM2.5 to PM10 concentrations. Moreover, Singh et al. (2013) found distinct seasonal patterns in the levels of PM10–2.5 across the DMR, ranging from highest during the dry summer months to lowest during the cold winter months. This is due to the greater load of coarser particles during the summer months as a result of dust transport form surrounding areas during the summer months compared to winter months.Table 1 Descriptive statistics for PM10 maximum, minimum, and average levels. Table 1Station Year Maximum Standard Deviation Minimum Standard Deviation Average Alipur 2019 862 127.924 30.25 14.1712 228.103 Alipur 2020 401.5 71.1405 30.75 10.8133 134.402 AshokVihar 2019 571 74.2654 35 10.7427 168.204 AshokVihar 2020 316 59.4934 17 5.89851 95.4233 AnandVihar 2019 929.75 156.419 53 25.1023 297.312 AnandVihar 2020 244.25 48.0845 26 16.5048 99.7517 Aya Nagar 2019 659.19 134.415 2.72 17.2113 155.674 Aya Nagar 2020 267.64 43.9914 2.5 8.94845 78.5316 Bawana 2019 879 135.246 19 20.1892 293.964 Bawana 2020 510 88.3488 40 13.8379 157.693 CRRI_Mathura 2019 722.3 110.78 9.57 24.3945 224.479 CRRI_Mathura 2020 522.22 82.9647 10.39 8.18575 105.91 DTU 2019 1000 205.991 18.75 17.3706 261.663 DTU 2020 359 64.773 23 8.67686 124.096 Karni Singh 2019 727 94.1505 7.5 11.7732 209.521 Karni Singh 2020 327 54.2852 21 7.14694 94.3364 Dwarka 2019 928 204.873 17.25 30.203 282.942 Dwarka 2020 488.5 87.5935 29 5.06922 116.574 IGI 2019 946.88 173.898 9.21 16.6022 224.754 IGI 2020 294.79 49.5147 2.73 9.04335 88.1071 ITO 2019 888 133.592 24 14.957 180.733 ITO 2020 336 76.5181 11 14.7504 92.4485 JhPuri 2019 757 90.972 17 21.3399 270.208 JhPuri 2020 374 69.8787 25 10.2574 126.443 JLN 2019 572.5 61.3752 12.5 17.7128 227.902 JLN 2020 350.75 59.2342 22.25 7.00038 95.8382 LodhiRoad 2019 693.88 116.096 11.53 12.9554 175.874 LodhiRoad 2020 348.12 85.1159 0.41 12.3119 87.5506 Major Dhyan 2019 498.75 56.9316 8.5 18.5752 200.816 Major Dhyan 2020 318.25 52.341 16 7.2127 89.018 Mandir Marg 2019 595 72.964 37 21.2579 229.153 Mandir Marg 2020 373.75 75.3458 16 8.43321 93.2862 Mundka 2019 973 124.8 19.5 21.7858 337.353 Mundka 2020 418.75 60.6591 23.25 7.09626 134.142 Najafgarh 2019 818.5 158.865 12.25 10.2112 191.994 Najafgarh 2020 614.5 101.876 18 16.3333 146.059 Narela 2019 721 88.8081 32.5 11.4617 277.344 Narela 2020 516 102.888 34.5 18.6923 157.114 North Campus 2019 852.95 113.615 8.74 23.7951 289.572 North Campus 2020 291.06 53.8835 0.3 11.1233 90.9651 Okhla 2019 875 120.866 17.5 11.9767 194.985 Okhla 2020 442 78.2785 27.5 6.75701 100.779 Patparganj 2019 723.75 113.481 62.5 23.5294 206.623 Patparganj 2020 389.25 71.8545 19 4.7531 81.7944 Punjabi Bagh 2019 675 90.6604 16.25 18.5927 223.43 Punjabi Bagh 2020 790.75 207.079 27.5 8.80379 112.304 PusaDPCC 2019 791 101.408 12.25 27.7238 221.084 PusaDPCC 2020 567 129.393 15.75 17.7111 88.3698 RKP 2019 850.75 120.056 17.75 20.104 235.803 RKP 2020 428 85.9955 11.25 11.5326 90.7663 Rohini 2019 781 122.662 13.75 17.467 268.78 Rohini 2020 566.75 92.7613 35 8.4801 137.409 SiriFort 2019 986.25 195.796 18 46.049 301.46 SiriFort 2020 563.25 105.185 25 6.1976 94.1084 SoniaVihar 2019 497 33.011 16 23.0697 233.819 SoniaVihar 2020 324 53.0742 25 7.43589 101.207 SriAurobindo 2019 748 116.038 4 16.5317 194.134 SriAurobindo 2020 238.5 35.6937 15.5 6.02407 71.6146 VivekVihar 2019 962 138.55 51 16.2336 245.17 VivekVihar 2020 428 85.9023 50 3.17364 107.087 Wazirpur 2019 907 109.841 31 30.9399 305.448 Wazirpur 2020 331 62.0834 24 9.41813 106.565 Table 2 Descriptive statistics for PM2.5 maximum, minimum, and average levels. Table 2Station Year Maximum Standard Deviation Minimum Standard Deviation Average Alipur 2019 447.5 108.302 8.75 4.55144 83.0603 Alipur 2020 149 38.1889 6.25 5.39188 47.5502 AshokVihar 2019 576.5 128.339 6 7.5641 95.3493 AshokVihar 2020 170 48.7817 5 5.67683 48.7782 Aya Nagar 2019 176.64 27.9336 0.13 7.65842 51.8133 Aya Nagar 2020 133.13 18.3072 0.16 3.08136 39.103 Bawana 2019 451 104.909 8.25 8.51754 103.071 Bawana 2020 279 66.855 8 6.92493 65.5673 CRRI_Mathura 2019 612.91 93.1076 0.08 6.39099 83.0624 CRRI_Mathura 2020 238.57 52.6926 0.82 5.58828 44.0428 DTU 2019 414.89 89.0331 12.37 5.69309 88.7776 DTU 2020 219.47 55.6613 3.62 5.80001 51.77 Karni Singh 2019 571 106.454 1 5.92663 65.8622 Karni Singh 2020 111.75 24.6156 2 5.78783 32.2407 Dwarka 2019 321.75 78.8187 3.75 7.60587 79.0549 Dwarka 2020 228.25 60.6285 5.75 4.69812 49.5281 IGI 2019 314.41 52.7348 2.24 5.73481 69.3283 IGI 2020 166.18 39.2024 0.6 3.40492 38.6598 IHBAS 2019 521.38 105.401 15 12.1524 111.183 IHBAS 2020 81.25 10.9892 10 0.15108 13.4519 ITO 2019 580 112.367 13 8.42679 88.8093 ITO 2020 329.5 84.6624 20 1.90715 74.5252 JhPuri 2019 806 143.623 1 9.89721 108.849 JhPuri 2020 184.25 46.0978 3.5 4.62052 53.7597 JLN 2019 294.25 65.4434 1.5 6.60945 65.0576 JLN 2020 179.25 42.377 1.25 4.89172 37.301 LodhiRoad 2019 220.37 37.1103 0.7 7.87626 56.9681 LodhiRoad 2020 157.77 35.5614 0.16 3.68312 46.4696 Major Dhyan 2019 741.5 141.564 8.25 7.25418 70.3656 Major Dhyan 2020 136.75 30.6355 3.5 4.1349 40.911 Mandir Marg 2019 341.25 55.2182 6.25 10.8533 76.6734 Mandir Marg 2020 153 31.0096 3.5 4.48487 36.8879 Mundka 2019 341 77.0833 4 7.28381 96.7941 Mundka 2020 276.25 68.01 4 5.59664 60.0536 NSIT 2019 607.36 134.288 6.57 7.24242 107.903 NSIT 2020 130.63 19.5396 21.96 4.09561 51.6994 Najafgarh 2019 268.5 68.7479 3 4.87394 68.406 Najafgarh 2020 191.5 57.9931 3.5 4.7637 47.9101 Narela 2019 542 113.72 10 6.85347 89.8291 Narela 2020 318.5 77.0327 4 12.0454 61.4989 North Campus 2019 381.93 83.2922 0.54 8.92598 93.7665 North Campus 2020 130.52 22.9001 0.24 1.94323 30.6803 Okhla 2019 288 65.7498 6.5 6.63966 73.1985 Okhla 2020 204 54.6407 7 4.95957 43.193 Patparganj 2019 320 71.1421 7 5.98171 67.463 Patparganj 2020 186.75 40.254 2.25 4.56474 36.5616 Punjabi Bagh 2019 328.5 75.4048 1 6.70212 73.3961 Punjabi Bagh 2020 715.25 205.564 2 4.98458 59.567 PusaDPCC 2019 301.5 62.3228 0.5 6.63007 61.599 PusaDPCC 2020 433.25 101.975 1 10.0395 41.749 RKP 2019 100 9.89105 3 19.6651 62.3571 RKP 2020 186 44.4902 1 6.07955 38.8317 Rohini 2019 523 118.413 7 7.25465 94.5361 Rohini 2020 376 74.5775 7 5.59098 60.8941 Shadipur 2019 754.18 176.328 7.62 10.9735 125.004 Shadipur 2020 88.75 15.9437 10 0.86196 18.426 SiriFort 2019 573 125.781 2.25 7.88153 76.1917 SiriFort 2020 459 83.9472 5 4.71554 42.7199 SoniaVihar 2019 383 89.404 5 10.0589 80.8302 SoniaVihar 2020 174.5 39.6314 5 5.06475 42.1439 SriAurobindo 2019 296.75 61.9354 1 5.94034 58.9903 SriAurobindo 2020 164.5 38.0119 3.5 4.24435 36.7326 VivekVihar 2019 592.75 146.715 2 9.02831 82.0469 VivekVihar 2020 228 61.9504 5 4.70387 45.9259 Wazirpur 2019 460.5 99.0444 12.5 10.2953 101.449 Wazirpur 2020 195.5 49.8522 7.5 4.98144 51.4931 In addition, we examined the change in monthly peak time of maximum for the two pollutants between the two years (Fig. 3 ). The peak time of maximum showed a gradual progression from after midnight to early morning hours for PM10 across most of the stations in a north to south direction during both years (Fig. 3a). In the case of PM2.5, the peak time of maximum occurrence occurred closer to the early morning hours, with a few hours earlier occurrence in the north relative to the south (Fig. 3b). The midnight to early morning maximum observed for both the pollutants can be attributed to the minimum variations in the convective available potential energy (CAPE) and other thermodynamic parameters in the early morning hours in the DMR (Ratnam et al., 2013). In addition there is relative lower atmospheric boundary layer and high traffic density in the early morning hours in the DMR. Similar results of higher levels of PM2.5 early morning and midnight were also found by Bhakta et al. (2019) from the analysis of 2 years of data at one station in the DMR. The results of their study also revealed a strong negative correlation between air temperatures and levels of PM2.5.Fig. 3 Spatial distribution of peak time of maximum (a) Average PM10; (b) Average PM2.5. The symbols pointing north indicate time of maximum at midnight, those pointing south indicate time of maximum at noon, and those pointing west indicate maximum at 6 p.m., and so on. Fig. 3 As seen in Fig. 3, the times of maximum PM10 or PM2.5 levels did not change appreciably between 2019 and 2020 at most stations. For each pollutant, the mean difference in the time of maximum across the station network was essentially equal to the standard error of the estimate in calculating the mean, thereby suggesting that the difference is not statistically significant. We also analyzed the change in the time of maximum for the monthly maximum and minimum levels, and the patterns were predominantly similar to that observed for the average monthly levels. The different sources of air pollution in the DMR are well documented, which include industrial activities, transport, road side construction, and regional emission sources that contribute a significant fraction to aerosol mass loading in the region (Nagpure et al., 2013; Saxena et al., 2014, 2017; Sen et al., 2016; Sharma et al., 2016). Thus, the almost complete stop in the rush hour traffic and other anthropogenic activities, including industrial and construction, can be considered as the major factors for the substantial decline in levels of particulate matter at the local level. It is also noteworthy that the spatial and temporal patterns of particulate matter in the DMR are a result of anthropogenic activities across the wider densely populated northern plains. The advective transport of particulate matter across the northern plains and consequent dispersal of pollutants to the marine atmospheric boundary layer of the Bay of Bengal has been well documented (Lelieveld et al., 2001; Sudheer & Sarin, 2008). Since the lockdown was at the national level, the results of our study can be representative of the levels of air pollution across the wider region of the Indian subcontinent. 4 Conclusions In the present study we have examined the impacts of lockdown in the DMR on the spatial patterns of air quality during April 2020. We analyzed two variables, PM2.5 and PM10 at the station level during April 2019 and 2020. The main findings of our study are summarized below:1. There was substantial decline in the levels of PM10 (20–70%) and PM2.5 (15–90%) in air quality across the DMR. 2. Spatially, the highest decline for the particulate matter was observed over the downtown core area and the adjacent industrial areas in the east and west. 3. The areas experiencing greater decline in the levels of particulate matter are associated with greater proportion of commercial land uses and economic activity in the form of offices and industries. Overall, the decline in PM10 was more widespread than PM2.5. 4. The diurnal patterns of the time of maximum for average monthly levels occurred closer to midnight for PM10 and early morning hours for PM2.5, which were in conformity with the results of previous studies. 5. The time of maximum values did not change significantly between 2019 and 2020. The results of our study highlight some of the positive impacts of the lockdown during a one month period on the local environment. As evident from the results of previously published studies, elevated levels of particulate matter in the DMR have led to increased rates of premature mortality in the DMR. An analysis of the relative contribution of various sectors to the levels of PM2.5 in the DMR and its surrounding area revealed transportation as the leading sector, followed by residential (in the form of wood, coal, kerosene, cow dung used with poor combustion technology in informal settlements, and liquefied petroleum gas with less emission in almost all houses), power plants (coal as fuel), and industrial sectors (Jain et al., 2018; Sahu et al., 2011). Therefore, with the implementation of a complete lockdown in the DMR leading to a steep decline in anthropogenic activities in the transportation and industrial sector resulted in the substantial improvement in air quality. Moreover, both of these pollutants have been consistently above the national standards and persistently represented a major challenge for policymakers and air quality scientists. However, it is noteworthy that the steep declines in levels of pollutants observed in different parts of the DMR have come at substantial social and economic costs, which make them difficult to sustain in the long-term. Further analysis is required to examine the implementation of similar phased lockdowns without excessive negative socio-economic impacts to achieve a more sustained decrease in levels of air pollution. This is particularly critical in view of the increased mortality, particularly an 11% increase in cardiovascular mortality as result of a 10 μgm−3 increase of PM2.5 (Bourdrel et al., 2017). Additionally, PM2.5 has been identified as the 5th risk factor of mortality, with 59% of those occurring in East and South Asia (Cohen et al., 2017). Therefore, it would be worthwhile to explore the impact of lower levels of particulate matter on the general health of the population in the DMR once the appropriate data are available for analysis. Author statement Shouraseni Sen Roy and Robert C. Balling Jr.: Conceptualization, Methodology, Data curation, Visualization, Investigation, Writing- Original draft preparation, Reviewing, and Editing. ==== Refs References Aneja V.P. Agarwal A. Roelle P.A. Phillips S.B. Tong Q.S. Watkins N. Yablonsky R. Measurements and analysis of criteria pollutants in New Delhi, India Environment International 27 2001 35 42 11488388 Balachandran S. Meena B.R. Khillare P.S. Particle size distribution and its elemental composition in the ambient air of Delhi Environment International 26 2000 49 54 11345737 Bhakta R. Khillare P.S. Jyethi D.S. Atmospheric particulate matter variations and comparison of two forecasting models for two Indian megacities Aerosol Science and Engineering 3 2 2019 54 62 Bourdrel T. Bind M.A. Béjot Y. Morel O. Argacha J.F. Cardiovascular effects of air pollution Archives of Cardiovascular Diseases 110 2017 634 642 28735838 Business Today Lockdown reduces Delhi pollution; air quality turns 'good' from 'hazardous.' Business Today https://www.businesstoday.in/latest/trends/lockdown--reduces-delhi-pollution-air-quality-turns-good-from-hazardous/story/399397.html 2020 Chowdhury S. Dey S. Cause-specific premature death from ambient PM2.5 exposure in India: Estimate adjusted for baseline mortality Environment International 91 2016 283 290 27063285 Clark P.J. Evans F.C. Distance to nearest neighbor as a measure of spatial relationships in populations Ecology 35 1954 445 453 Cohen A.J. Brauer M. Burnett R. Estimates and 25-year trends of the global burden of disease attributable to ambient air pollution: An analysis of data from the global burden of diseases study 2015 The Lancet 389 2017 1907 1918 Ellis-Petersen H. Ratcliffe R. Cowie S. Daniels J.P. Kuo L. It's positively alpine!: Disbelief in big cities as air pollution falls. The Guardian https://www.theguardian.com/environment/2020/apr/11/positively-alpine-disbelief-air-pollution-falls-lockdown-coronavirus 2020 Ghude S.D. Chate D.M. Jena C. Beig G. Kumar R. Barth M.C. Pfister G.G. Fadnavis S. Pithani P. Premature mortality in India due to PM2.5 and ozone exposure Geophysical Research Letters 43 2016 4650 4658 Gurjar B.R. Ravindra K. Nagpure A.S. Air pollution trends over India megacities and their local-to-global implications Atmospheric Environment 142 2016 475 495 Gurjar B.R. van Aardenne J.A. Lelieveld J. Mohan M. Emission estimates and trends (1990–2000) for mega city Delhi and implications Atmospheric Environment 38 2004 5663 5681 Jain M. Dawa D. Mehta R. Dimri A.P. Pandit M.K. Monitoring land use change and its drivers in Delhi, India using multi-temporal satellite data Modeling Earth Systems and Environment 2 1 2016 19 Jain S. Sharma S.K. Mandal T.K. Saxena M. Source apportionment of PM10 in Delhi, India using PCA/APCS, UNMIX and PMF Particuology 37 2018 107 118 Kandlikar M. Air pollution at a hotspot location in Delhi: Detecting trends, seasonal cycles and oscillations Atmospheric Environment 41 2007 5934 5947 Kumari P. Toshniwal D. Impact of lockdown measures during COVID-19 on air quality–A case study of India International Journal of Environmental Health Research 2020 1 8 Lelieveld J.O. Crutzen P.J. Ramanathan V. Andreae M.O. Brenninkmeijer C.A.M. Campos T. Cass G.R. Dickerson R.R. Fischer H. De Gouw J.A. Hansel A. The Indian ocean experiment: Widespread air pollution from south and southeast Asia Science 291 5506 2001 1031 1036 11161214 Liu Z. Sen Roy S. Spatial patterns of seasonal level diurnal variations of ozone and respirable suspended particulates in Hong Kong The Professional Geographer 67 1 2014 17 27 10.1080/00330124.2014.886922 Mahato S. Pal S. Ghosh K.G. Effect of lockdown amid COVID-19 pandemic on air quality of the megacity Delhi, India The Science of the Total Environment 2020 139086 Nagar P.K. Sharma M. Das D. A new method for trend analyses in PM10 and impact of crop residue burning in Delhi, Kanpur and Jaipur, India Urban Climate 27 2019 193 203 Nagpure A.S. Sharma K. Gurjar B.R. Traffic induced emission estimates and trends (2000–2005) in megacity Delhi Urban Climate 4 2013 61 73 Nelson H. Harmonic analysis 1983 Addison-Wesley Reading, Massachusetts Oliver M.A. Webster R. Kriging: A method of interpolation for geographical information systems International Journal of Geographical Information Systems 4 1990 313 332 O'Shea P.M. Sen Roy S. Singh R.B. Diurnal variations in the spatial patterns of air pollution across the Delhi metropolitan region Theoretical and Applied Climatology 124 3 2015 609 620 10.1007/s00704-015-1441-y Patel K. Airborne particle levels plummet in northern India https://earthobservatory.nasa.gov/images/146596/airborne-particle-levels-plummet-in-northern-india 2020 Ratnam M.V. Santhi Y.D. Rajeevan M. Rao S.V.B. Diurnal variability of stability indices observed using radiosonde observations over a tropical station: Comparison with microwave radiometer measurements Atmospheric Research 124 2013 21 33 Sahu S.K. Beig G. Parkhi N.S. Emissions inventory of anthropogenic PM2. 5 and PM10 in Delhi during commonwealth games 2010 Atmospheric Environment 45 34 2011 6180 6190 Saxena M. Sharma A. Sen A. Saxena P. Mandal T.K. Sharma S.K. Sharma C. Water soluble inorganic species of PM10 and PM2. 5 at an urban site of Delhi, India: Seasonal variability and sources Atmospheric Research 184 2017 112 125 Saxena M. Singh D.P. Saud T. Gadi R. Singh S. Sharma S. Mandal T.K. Study on particulate polycyclic aromatic hydrocarbons over Bay of Bengal Atmospheric Research 145– 146 2014 205 217 Sen Roy S. Impact of lunar cycle on the precipitation in India Geophysical Research Letters 33 1 2006 Sen Roy S. The impacts of ENSO, PDO, and local SSTs on winter precipitation in India Physical Geography 27 2006 464 474 Sen Roy S. Rahman A. Ahmed S. Shahfahad Ahmad I.A. Alarming groundwater depletion in the Delhi metropolitan region: A long-term assessment Environmental Monitoring and Assessment 192 2020 10.1007/s10661-020-08585-8 Sen A. Ahammed Y.N. Banerjee T. Chatterjee A. Choudhuri A.K. Das T. Deb N.C. Dhir A. Goel S. Khan A.H. Mandal T.K. Murari V. Rao P.S. Saxena M. Sharma S.K. Sharma A. Vachaspati C.V. Spatial variability in ambient atmospheric fine and coarse mode aerosols over indo-Gangetic plains, India and adjoining oceans during the onset of summer monsoons Atmos. Poll. Res. 7 2016 521 532 Sharma S.K. Mandal T.K. Jain S. Saraswati S.A. Saxena M. Source apportionment of PM2.5 in Delhi, India using PMF model Bulletin of Environmental Contamination and Toxicology 97 2016 286 293 27209541 Shehzad K. Sarfraz M. Shah S.G.M. The impact of COVID-19 as a necessary evil on air pollution in India during the lockdown Environmental Pollution 266 2020 115080 32634726 Singh V. Singh S. Biswal A. Kesarkar A.P. Mor S. Ravindra K. Diurnal and temporal changes in air pollution during COVID-19 strict lockdown over different regions of India Environmental Pollution 266 2020 115368 32829030 Singh K. Tiwari S. Jha A.K. Aggarwal S.G. Bisht D.S. Murty B.P. Khan Z.H. Gupta P.K. Mass-size distribution of PM 10 and its characterization of ionic species in fine (PM 2.5) and coarse (PM 10− 2.5) mode, New Delhi, India Natural Hazards 68 2 2013 775 789 Sudheer A.K. Sarin M.M. Carbonaceous aerosols in MABL of Bay of bengal: Influence of continental outflow Atmospheric Environment 42 18 2008 4089 4100 Tiwari S. Hopke P.K. Pipal A.S. Srivastava A.K. Bisht D.S. Tiwari S. Singh A.K. Soni V.K. Attri S.D. Intra-urban variability of particulate matter (PM2. 5 and PM10) and its relationship with optical properties of aerosols over Delhi, India Atmospheric Research 166 2015 223 232 Venter Z.S. Aunan K. Chowdhury S. Lelieveld J. COVID-19 lockdowns cause global air pollution declines with implications for public health risk https://www.medrxiv.org/content/10.1101/2020.04.10.20060673v1.article-metrics 2020 Wu X. Nethery R.C. Sabath B.M. Braun D. Dominici F. Exposure to air pollution and COVID-19 mortality in the United States 2020 medRxiv 2020

0

PMC9753935

NO-CC CODE

2022-12-16 23:26:22

no

Appl Geogr. 2021 Mar 1; 128:102418

utf-8

Appl Geogr

10.1016/j.apgeog.2021.102418

oa_other

==== Front Appl Geogr Appl Geogr Applied Geography (Sevenoaks, England) 0143-6228 0143-6228 Elsevier Ltd. S0143-6228(21)00034-5 10.1016/j.apgeog.2021.102418 102418 Article Impact of the COVID-19 lockdown on air quality in the Delhi Metropolitan Region Roy Shouraseni Sen a∗ Balling Robert C. Jr. b a Department of Geography and Regional Studies, University of Miami, FL, USA b School of Geographical Sciences and Urban Planning, Arizona State University, AZ, USA ∗ Corresponding author. 1 3 2021 3 2021 1 3 2021 128 102418102418 10 6 2020 4 1 2021 27 1 2021 © 2021 Elsevier Ltd. All rights reserved. 2021 Elsevier Ltd Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. With the rapid spread of COVID-19 related cases globally, national governments took different lockdown approaches to limit the spread of the virus. Among them, the Government of India imposed a complete nationwide lockdown starting on March 25, 2020. This presented a unique opportunity to explore how a complete standstill in regular daily activities might impact the local environment. In this study, we have analyzed the change in the air quality levels stemming from the reduced anthropogenic activities in one of the most polluted cities in the world, the Delhi Metropolitan Region (DMR). We analyzed station-level changes in particulate matter, PM10 and PM2.5, across the DMR between April 2019 and 2020. The results of our study showed widespread reduction in the levels of both pollutants, with substantial spatial variations. The largest decreases in particulate matter were associated with industrial and commercial areas. Highest levels of PM10 and PM2.5 were observed near sunrise with little change in the time of maximum between 2019 and 2020. The results of our study highlight the role of anthropogenic activities on the air quality at the local level. Keywords COVID-19 PM10 PM2.5 Delhi metropolitan region Harmonics Diurnal ==== Body pmc1 Introduction Within a few months in 2020, COVID-19 had massively impacted the entire globe far beyond the direct effects on human health. These impacts vary at different spatial scales as reflected in the number of infections and mortality, as well as the indirect impacts on the economy and environment. Many local and national governments implemented various degrees of lockdown or “shelter at home” policies to contain the spread of this global pandemic. This has resulted in a massive reduction in global economic activity thereby significantly lowering air pollution levels in many areas of the world including China (Wu et al., 2020) and India (Patel et al., 2020). Specifically, substantial declines in ground level nitrogen dioxide, ozone, and particulate matter were reported within the first two weeks of the lockdown in 27 countries as revealed from the analysis of satellite data during February and March 2020 (Venter et al., 2020). However, due to the coarse spatial and temporal resolution of the satellite data, their study was limited in revealing the local level patterns in air pollution. This is particularly relevant in parts of Asia, such as in India, where most of the high pollution levels are concentrated in the large urban areas. Therefore, in the present study we have analyzed the change in levels of two pollutants, PM10 and PM2.5, over a one month period between April 2019 and 2020 in the Delhi Metropolitan Region (DMR), India (Fig. 1 ).Fig. 1 Distribution of air pollution monitoring stations in the Delhi Metropolitan Region (DMR). Fig. 1 The Government of India imposed lockdown on all of its 1.3 billion citizens on March 25, 2020 to limit the spread of the COVID-19; the lockdown ended after 55 days on May 19, 2020. As a result, there was widespread shutdown of factories, road, and air traffic across the DMR. The shutdown was so severe that it triggered a massive migration of wage laborers walking for several 100s of kilometers across northern India to their villages caused by no available jobs and the cancellation of long distance transportation such as railways and bus systems. The initial results from satellite image analysis of aerosol optical depths revealed a significant drop across the northern plains resulting from the closure of coal-fired heavy industries and reduction in traffic across large urban areas like the DMR (Patel et al., 2020). In addition, there have been widespread reports in mainstream media outlets about the reduction in pollution levels in the DMR dropping from unhealthy and hazardous to good (Business Today, 2020; Ellis-Peterson et al., 2020). These trends have also been documented in recently published studies. For instance, Shehzad et al. (2020) analyzed Sentinel – 5 P satellite images and reported a significant improvement in air quality and a 40–50% decrease in atmospheric nitrogen dioxide levels in the large cities of India, including DMR and Mumbai. More detailed analysis of station level data for the DMR during pre and post lockdown periods revealed significant decreases in levels of particulate matter within days of the lockdown (Kumari & Toshniwal, 2020; Mahato et al., 2020). Additionally, detailed hourly analysis of air pollutants revealed a decline in the levels of particulate matter during nighttime and peak traffic hours (Singh et al., 2020). This is particularly significant in view of the DMR and its suburbs being ranked as the most polluted city in the world for at least the last five years. Specifically, the pollutants with consistently high levels in the DMR are PM10, PM2.5, NO2, SO2, and O3 (Aneja et al., 2001; Balachandran et al., 2000; Gurjar et al., 2004; Kandikar 2007; Nagar et al., 2019; O'Shea et al., 2015; Sahu et al., 2011). Furthermore, India has recorded an increasing trend in population-weighted mean concentrations of PM2.5, with a noticeable increase since 2010 (Bhakta et al., 2019; Gurjar et al., 2016). These elevated concentrations of PM2.5 exposures have been attributed to annual premature death estimates of 272,000 for chronic obstructive pulmonary disease (COPD), 110,600 for ischemic heart disease, and 14,800 for lung cancer (Chowdhury & Dey, 2016). Specifically, for the DMR a reduction in life expectancy due to exposure to PM2.5 was 6.3 ± 2.2 years greater than the same for overall Indian population (3.4 ± 1.1 years) (Ghude et al., 2016). Due to the paucity of data, most of the above mentioned station level studies in the DMR have examined the variations for select pollutants in limited number of stations, ranging between two and five. The results from these studies revealed significant variations in the levels of various pollutants and diurnal cycles across the study area. All of these studies highlighted the excessive readings and increase in the levels of pollutants in the DMR, mainly resulting from traffic congestion and industrial activities. Thus, with the implementation of the strict measures associated with the lockdown, the drop in air pollution levels is distinctly visible and merits detailed analysis. Therefore, the two specific objectives of the study are:1. Determine the local-level spatial patterns of change in the levels of two major pollutants, PM10 and PM2.5, between April 2019 and 2020. 2. Analyze the changes in the time of maximum for peak levels in PM10 and PM2.5 between April 2019 and 2020. 2 Data and methods Station-level hourly data for PM10 and PM2.5 were obtained from the Central Pollution Control Board (CPCB), collected as part of the Air Quality Monitoring Program in India. The data were collected for April 2019 and 2020 to assess the change in pollution levels due to the lockdown. There are 40 pollution monitoring stations spread across the DMR, out of which there were continuous data available for 34 stations for PM10 and 31 stations for PM2.5 (Fig. 1). We analyzed only two pollutants PM10 and PM2.5 because these were the only two pollutants with complete data across all the stations. The data for the other pollutants were incomplete and thus not suitable for analysis. Furthermore, these two pollutants have consistently been the biggest challenge for air quality scientists and policymakers and therefore the results of the study are particularly relevant. In order to test the spatial distribution of the station network across the study area, the nearest-neighbor statistics were calculated as the ratio between the observed mean distance among the station locations and the expected mean distance given a random distribution (Clark & Evans, 1954). The ratios of 1.03 for PM10 and 0.90 for PM2.5 show that both station networks have a random distribution. We analyzed the changes in the levels of PM10 and PM2.5 for the monthly average values by calculating the differences at the station level. The percentage change in the levels of the pollutants at the station level were visualized using spatial interpolation, simple kriging. Kriging is a surface interpolation method utilized to visualize spatial variation through a variogram, thus minimizing prediction errors (Oliver & Webster, 1990; Sen Roy, 2006a). The final interpolated surface is calculated by incorporating the spatial and statistical relationships among the different variables by using the following equation:Z(s) = μ + ε(s) Where μ is a known constant utilized to interpolate the resulting surface, s denotes the location being predicted, and ε(s) is the error term (Sen Roy, 2006b). This method was preferred because of its accuracy in surface interpolation and lower root mean square error. Harmonic analysis was used to determine the time of maximum concentration of PM10 and PM2.5 levels based on the average hourly values over the course of a month. To fit a trigonometric wave with one maxima and one minima, the harmonic equation for any station with 24 values takes the form:f(x)=X‾+∑r=1N/2Arcos(rθ−Φr) where f(x) is the estimated value in each interval, X‾ is the average value over the N = 24 intervals, A r is the amplitude of the rth harmonic wave, r is the frequency or number of times the harmonic wave is repeated over the fundamental period (in this case r = 1), θ is derived as 2πx/N where x signifies the intervals over the fundamental period, and Φris the phase angle of the rth harmonic reinterpreted as the time of maximum. The basic form is explained below:f(x)=X‾+∑r=1N/2[arsin(2πrx/P)+brcos(2πrx/P)] where the Fourier coefficients, a r and b r, are calculated as:ar=∑x=1N2N[f(x)sin(2πrx/P)] andbr=∑x=1N2N[f(x)cos(2πrx/P)] The amplitude, A r, is calculated as (a r 2 + b r 2 ) 0.5, the phase angle, Φr, equals tan−1 (a r /b r), and the portion of variance denoted by the rth harmonic wave, V r, is calculated as A r 2 /2s 2 wherein s is the standard deviation of the N values (see Nelson, 1983, p. 190). Given that the harmonic wave is fitted to PM10 or PM2.5 averages in 24 hourly intervals, the explained variance levels had to be > 0.16 to be statistically significant at the 0.05 level of confidence. This method has been successfully used to study diurnal patterns in air pollution variables in previous studies (Liu & Sen Roy, 2014; O'Shea et al., 2015). 3 Results and discussion The DMR is located in the northern interior of the Indian subcontinent, and thus experiences a typical continental climate. The comparative results from the two different years reveal substantially lower levels of in the levels of air pollution between the two years with distinct spatial variations across the DMR (Fig. 2 ). Typical of megacities of Asia, the DMR is densely populated (greater than 25,000 persons per km2), with an estimated total population of about 17 million. Majority of the DMR is urban, with 97.5% of the population identified as urban (Sen Roy et al., 2020). According to Jain et al. (2016), the net percent change in land use from 1977 to 2014 for urban built-up areas increased by 30.61%, along with a decrease by 22.75% for cultivated areas, 5.31% for dense forest, and 2.76% for wasteland. With the steep increase in population accompanied by car ownership and unplanned urbanization, the DMR has experienced steep increase in levels of air pollution over the years. The extremely hazardous levels of air pollution in the winter months has led to the forced closure of schools and steep increase respiratory diseases among the local population. The sources of air pollution are not only from local sources such as vehicular emissions and industrial activities, but more recently the burning of crops after harvesting in the agricultural fields in the neighboring states of Punjab and Haryana. Thus, the pollution levels noted in 2019 typifies that time of the year in the DMR over the last decade.Fig. 2 Changes in average hourly levels of particulate matter between April 2019 and 2020 (a) PM10; (b) PM2.5. Fig. 2 As is evident from Fig. 2, the level for both pollutants in 2019 were higher than those observed in 2020 for all the stations. The percentage change for PM10 ranged between 20 and 70%, while the range of decline was wider for PM2.5 at 15–90%. At the local level, 21 out of 31 stations experienced greater than 50% decline for PM10. The largest declines were concentrated in the eastern half of the DMR, which also consistently experienced higher levels of particulate matter. Moreover, for the PM2.5 the largest percentage declines (greater than 50%) were observed for only 8 out of 34 stations located in the central and eastern parts of the DMR. The greater decline in central DMR is due to the lower levels of economic activity associated with the high density of office buildings and commercial areas in the core downtown area. Similarly, the higher rates of decline in the eastern DMR are associated with the closure of local factories and thermal power plants. Specifically, there are three thermal power plants located in the DMR, which include Indraprastha Power Station and Rajghat Power House in the east, and Badarpur Power Station in the northwest. Majority of the land use in the northern and western DMR is residential, and thus experienced relatively lower levels of decline in the pollutants. Two stations, located in Shadipur (west of central downtown) and Dilshad Garden (east) experienced greater than 85% decline in PM2.5 (Fig. 2b and Table 2 ). The lowest differences in average hourly levels of PM10 were located outside the central core of the DMR, such as Najafgarh (192 μg/m3 in 2019 vs 146.1ug/m3 in 2020), Ashok Vihar (168.2ug/m3 in 2019 vs 95.42μg/m3 in 2020), and Aya Nagar (155.7ug/m3 in 2019 vs 78.53μg/m3 in 2020) (Table 1, Fig. 2a). Among these three stations, two of them are predominantly residential, while Najafgarh consists of transitioning from rural to urban land uses mixed with industries. Similarly in the case of PM2.5, the lowest differences in average hourly levels were observed in predominantly residential areas, including Lodhi Road (57μg/m3 in 2019 vs 46.5ug/m3 in 2020), Aya Nagar (51.8ug/m3 in 2019 vs 39.1ug/m3 in 2020), and Punjabi Bagh (155.7ug/m3 in 2019 vs 73.4ug/m3 in 2020) (Table 2, Fig. 2b). Moreover, the hourly maximum values between the two years revealed higher values across most of the stations during 2019 for both PM10 and PM2.5, except in Punjabi Bagh (for both pollutants) and Pusa DPCC (PM2.5) in West Delhi (Table 1, Table 2). There was also a greater amount of variance in the hourly maximum levels for both of the pollutants during 2019 compared to 2020. However, the patterns were not as distinct in the case of hourly minimum values across the DMR. Our results are in conformity with the results of a previous study by Tiwari et al. (2015), who found substantial spatial variations in the correlation between observed levels of PM10 and PM2.5 at the seasonal scale and weekday vs weekends for a limited number of stations. Specifically, their analysis of hourly level observations of these two particulates revealed the mean coarse mode particulate mass concentration (PM10–2.5) as 113.6 ± 70.4 μg/m3 (varied from 13.6 to 630 μg/m3) constituting about 49% contribution of PM2.5 to PM10 concentrations. Moreover, Singh et al. (2013) found distinct seasonal patterns in the levels of PM10–2.5 across the DMR, ranging from highest during the dry summer months to lowest during the cold winter months. This is due to the greater load of coarser particles during the summer months as a result of dust transport form surrounding areas during the summer months compared to winter months.Table 1 Descriptive statistics for PM10 maximum, minimum, and average levels. Table 1Station Year Maximum Standard Deviation Minimum Standard Deviation Average Alipur 2019 862 127.924 30.25 14.1712 228.103 Alipur 2020 401.5 71.1405 30.75 10.8133 134.402 AshokVihar 2019 571 74.2654 35 10.7427 168.204 AshokVihar 2020 316 59.4934 17 5.89851 95.4233 AnandVihar 2019 929.75 156.419 53 25.1023 297.312 AnandVihar 2020 244.25 48.0845 26 16.5048 99.7517 Aya Nagar 2019 659.19 134.415 2.72 17.2113 155.674 Aya Nagar 2020 267.64 43.9914 2.5 8.94845 78.5316 Bawana 2019 879 135.246 19 20.1892 293.964 Bawana 2020 510 88.3488 40 13.8379 157.693 CRRI_Mathura 2019 722.3 110.78 9.57 24.3945 224.479 CRRI_Mathura 2020 522.22 82.9647 10.39 8.18575 105.91 DTU 2019 1000 205.991 18.75 17.3706 261.663 DTU 2020 359 64.773 23 8.67686 124.096 Karni Singh 2019 727 94.1505 7.5 11.7732 209.521 Karni Singh 2020 327 54.2852 21 7.14694 94.3364 Dwarka 2019 928 204.873 17.25 30.203 282.942 Dwarka 2020 488.5 87.5935 29 5.06922 116.574 IGI 2019 946.88 173.898 9.21 16.6022 224.754 IGI 2020 294.79 49.5147 2.73 9.04335 88.1071 ITO 2019 888 133.592 24 14.957 180.733 ITO 2020 336 76.5181 11 14.7504 92.4485 JhPuri 2019 757 90.972 17 21.3399 270.208 JhPuri 2020 374 69.8787 25 10.2574 126.443 JLN 2019 572.5 61.3752 12.5 17.7128 227.902 JLN 2020 350.75 59.2342 22.25 7.00038 95.8382 LodhiRoad 2019 693.88 116.096 11.53 12.9554 175.874 LodhiRoad 2020 348.12 85.1159 0.41 12.3119 87.5506 Major Dhyan 2019 498.75 56.9316 8.5 18.5752 200.816 Major Dhyan 2020 318.25 52.341 16 7.2127 89.018 Mandir Marg 2019 595 72.964 37 21.2579 229.153 Mandir Marg 2020 373.75 75.3458 16 8.43321 93.2862 Mundka 2019 973 124.8 19.5 21.7858 337.353 Mundka 2020 418.75 60.6591 23.25 7.09626 134.142 Najafgarh 2019 818.5 158.865 12.25 10.2112 191.994 Najafgarh 2020 614.5 101.876 18 16.3333 146.059 Narela 2019 721 88.8081 32.5 11.4617 277.344 Narela 2020 516 102.888 34.5 18.6923 157.114 North Campus 2019 852.95 113.615 8.74 23.7951 289.572 North Campus 2020 291.06 53.8835 0.3 11.1233 90.9651 Okhla 2019 875 120.866 17.5 11.9767 194.985 Okhla 2020 442 78.2785 27.5 6.75701 100.779 Patparganj 2019 723.75 113.481 62.5 23.5294 206.623 Patparganj 2020 389.25 71.8545 19 4.7531 81.7944 Punjabi Bagh 2019 675 90.6604 16.25 18.5927 223.43 Punjabi Bagh 2020 790.75 207.079 27.5 8.80379 112.304 PusaDPCC 2019 791 101.408 12.25 27.7238 221.084 PusaDPCC 2020 567 129.393 15.75 17.7111 88.3698 RKP 2019 850.75 120.056 17.75 20.104 235.803 RKP 2020 428 85.9955 11.25 11.5326 90.7663 Rohini 2019 781 122.662 13.75 17.467 268.78 Rohini 2020 566.75 92.7613 35 8.4801 137.409 SiriFort 2019 986.25 195.796 18 46.049 301.46 SiriFort 2020 563.25 105.185 25 6.1976 94.1084 SoniaVihar 2019 497 33.011 16 23.0697 233.819 SoniaVihar 2020 324 53.0742 25 7.43589 101.207 SriAurobindo 2019 748 116.038 4 16.5317 194.134 SriAurobindo 2020 238.5 35.6937 15.5 6.02407 71.6146 VivekVihar 2019 962 138.55 51 16.2336 245.17 VivekVihar 2020 428 85.9023 50 3.17364 107.087 Wazirpur 2019 907 109.841 31 30.9399 305.448 Wazirpur 2020 331 62.0834 24 9.41813 106.565 Table 2 Descriptive statistics for PM2.5 maximum, minimum, and average levels. Table 2Station Year Maximum Standard Deviation Minimum Standard Deviation Average Alipur 2019 447.5 108.302 8.75 4.55144 83.0603 Alipur 2020 149 38.1889 6.25 5.39188 47.5502 AshokVihar 2019 576.5 128.339 6 7.5641 95.3493 AshokVihar 2020 170 48.7817 5 5.67683 48.7782 Aya Nagar 2019 176.64 27.9336 0.13 7.65842 51.8133 Aya Nagar 2020 133.13 18.3072 0.16 3.08136 39.103 Bawana 2019 451 104.909 8.25 8.51754 103.071 Bawana 2020 279 66.855 8 6.92493 65.5673 CRRI_Mathura 2019 612.91 93.1076 0.08 6.39099 83.0624 CRRI_Mathura 2020 238.57 52.6926 0.82 5.58828 44.0428 DTU 2019 414.89 89.0331 12.37 5.69309 88.7776 DTU 2020 219.47 55.6613 3.62 5.80001 51.77 Karni Singh 2019 571 106.454 1 5.92663 65.8622 Karni Singh 2020 111.75 24.6156 2 5.78783 32.2407 Dwarka 2019 321.75 78.8187 3.75 7.60587 79.0549 Dwarka 2020 228.25 60.6285 5.75 4.69812 49.5281 IGI 2019 314.41 52.7348 2.24 5.73481 69.3283 IGI 2020 166.18 39.2024 0.6 3.40492 38.6598 IHBAS 2019 521.38 105.401 15 12.1524 111.183 IHBAS 2020 81.25 10.9892 10 0.15108 13.4519 ITO 2019 580 112.367 13 8.42679 88.8093 ITO 2020 329.5 84.6624 20 1.90715 74.5252 JhPuri 2019 806 143.623 1 9.89721 108.849 JhPuri 2020 184.25 46.0978 3.5 4.62052 53.7597 JLN 2019 294.25 65.4434 1.5 6.60945 65.0576 JLN 2020 179.25 42.377 1.25 4.89172 37.301 LodhiRoad 2019 220.37 37.1103 0.7 7.87626 56.9681 LodhiRoad 2020 157.77 35.5614 0.16 3.68312 46.4696 Major Dhyan 2019 741.5 141.564 8.25 7.25418 70.3656 Major Dhyan 2020 136.75 30.6355 3.5 4.1349 40.911 Mandir Marg 2019 341.25 55.2182 6.25 10.8533 76.6734 Mandir Marg 2020 153 31.0096 3.5 4.48487 36.8879 Mundka 2019 341 77.0833 4 7.28381 96.7941 Mundka 2020 276.25 68.01 4 5.59664 60.0536 NSIT 2019 607.36 134.288 6.57 7.24242 107.903 NSIT 2020 130.63 19.5396 21.96 4.09561 51.6994 Najafgarh 2019 268.5 68.7479 3 4.87394 68.406 Najafgarh 2020 191.5 57.9931 3.5 4.7637 47.9101 Narela 2019 542 113.72 10 6.85347 89.8291 Narela 2020 318.5 77.0327 4 12.0454 61.4989 North Campus 2019 381.93 83.2922 0.54 8.92598 93.7665 North Campus 2020 130.52 22.9001 0.24 1.94323 30.6803 Okhla 2019 288 65.7498 6.5 6.63966 73.1985 Okhla 2020 204 54.6407 7 4.95957 43.193 Patparganj 2019 320 71.1421 7 5.98171 67.463 Patparganj 2020 186.75 40.254 2.25 4.56474 36.5616 Punjabi Bagh 2019 328.5 75.4048 1 6.70212 73.3961 Punjabi Bagh 2020 715.25 205.564 2 4.98458 59.567 PusaDPCC 2019 301.5 62.3228 0.5 6.63007 61.599 PusaDPCC 2020 433.25 101.975 1 10.0395 41.749 RKP 2019 100 9.89105 3 19.6651 62.3571 RKP 2020 186 44.4902 1 6.07955 38.8317 Rohini 2019 523 118.413 7 7.25465 94.5361 Rohini 2020 376 74.5775 7 5.59098 60.8941 Shadipur 2019 754.18 176.328 7.62 10.9735 125.004 Shadipur 2020 88.75 15.9437 10 0.86196 18.426 SiriFort 2019 573 125.781 2.25 7.88153 76.1917 SiriFort 2020 459 83.9472 5 4.71554 42.7199 SoniaVihar 2019 383 89.404 5 10.0589 80.8302 SoniaVihar 2020 174.5 39.6314 5 5.06475 42.1439 SriAurobindo 2019 296.75 61.9354 1 5.94034 58.9903 SriAurobindo 2020 164.5 38.0119 3.5 4.24435 36.7326 VivekVihar 2019 592.75 146.715 2 9.02831 82.0469 VivekVihar 2020 228 61.9504 5 4.70387 45.9259 Wazirpur 2019 460.5 99.0444 12.5 10.2953 101.449 Wazirpur 2020 195.5 49.8522 7.5 4.98144 51.4931 In addition, we examined the change in monthly peak time of maximum for the two pollutants between the two years (Fig. 3 ). The peak time of maximum showed a gradual progression from after midnight to early morning hours for PM10 across most of the stations in a north to south direction during both years (Fig. 3a). In the case of PM2.5, the peak time of maximum occurrence occurred closer to the early morning hours, with a few hours earlier occurrence in the north relative to the south (Fig. 3b). The midnight to early morning maximum observed for both the pollutants can be attributed to the minimum variations in the convective available potential energy (CAPE) and other thermodynamic parameters in the early morning hours in the DMR (Ratnam et al., 2013). In addition there is relative lower atmospheric boundary layer and high traffic density in the early morning hours in the DMR. Similar results of higher levels of PM2.5 early morning and midnight were also found by Bhakta et al. (2019) from the analysis of 2 years of data at one station in the DMR. The results of their study also revealed a strong negative correlation between air temperatures and levels of PM2.5.Fig. 3 Spatial distribution of peak time of maximum (a) Average PM10; (b) Average PM2.5. The symbols pointing north indicate time of maximum at midnight, those pointing south indicate time of maximum at noon, and those pointing west indicate maximum at 6 p.m., and so on. Fig. 3 As seen in Fig. 3, the times of maximum PM10 or PM2.5 levels did not change appreciably between 2019 and 2020 at most stations. For each pollutant, the mean difference in the time of maximum across the station network was essentially equal to the standard error of the estimate in calculating the mean, thereby suggesting that the difference is not statistically significant. We also analyzed the change in the time of maximum for the monthly maximum and minimum levels, and the patterns were predominantly similar to that observed for the average monthly levels. The different sources of air pollution in the DMR are well documented, which include industrial activities, transport, road side construction, and regional emission sources that contribute a significant fraction to aerosol mass loading in the region (Nagpure et al., 2013; Saxena et al., 2014, 2017; Sen et al., 2016; Sharma et al., 2016). Thus, the almost complete stop in the rush hour traffic and other anthropogenic activities, including industrial and construction, can be considered as the major factors for the substantial decline in levels of particulate matter at the local level. It is also noteworthy that the spatial and temporal patterns of particulate matter in the DMR are a result of anthropogenic activities across the wider densely populated northern plains. The advective transport of particulate matter across the northern plains and consequent dispersal of pollutants to the marine atmospheric boundary layer of the Bay of Bengal has been well documented (Lelieveld et al., 2001; Sudheer & Sarin, 2008). Since the lockdown was at the national level, the results of our study can be representative of the levels of air pollution across the wider region of the Indian subcontinent. 4 Conclusions In the present study we have examined the impacts of lockdown in the DMR on the spatial patterns of air quality during April 2020. We analyzed two variables, PM2.5 and PM10 at the station level during April 2019 and 2020. The main findings of our study are summarized below:1. There was substantial decline in the levels of PM10 (20–70%) and PM2.5 (15–90%) in air quality across the DMR. 2. Spatially, the highest decline for the particulate matter was observed over the downtown core area and the adjacent industrial areas in the east and west. 3. The areas experiencing greater decline in the levels of particulate matter are associated with greater proportion of commercial land uses and economic activity in the form of offices and industries. Overall, the decline in PM10 was more widespread than PM2.5. 4. The diurnal patterns of the time of maximum for average monthly levels occurred closer to midnight for PM10 and early morning hours for PM2.5, which were in conformity with the results of previous studies. 5. The time of maximum values did not change significantly between 2019 and 2020. The results of our study highlight some of the positive impacts of the lockdown during a one month period on the local environment. As evident from the results of previously published studies, elevated levels of particulate matter in the DMR have led to increased rates of premature mortality in the DMR. An analysis of the relative contribution of various sectors to the levels of PM2.5 in the DMR and its surrounding area revealed transportation as the leading sector, followed by residential (in the form of wood, coal, kerosene, cow dung used with poor combustion technology in informal settlements, and liquefied petroleum gas with less emission in almost all houses), power plants (coal as fuel), and industrial sectors (Jain et al., 2018; Sahu et al., 2011). Therefore, with the implementation of a complete lockdown in the DMR leading to a steep decline in anthropogenic activities in the transportation and industrial sector resulted in the substantial improvement in air quality. Moreover, both of these pollutants have been consistently above the national standards and persistently represented a major challenge for policymakers and air quality scientists. However, it is noteworthy that the steep declines in levels of pollutants observed in different parts of the DMR have come at substantial social and economic costs, which make them difficult to sustain in the long-term. Further analysis is required to examine the implementation of similar phased lockdowns without excessive negative socio-economic impacts to achieve a more sustained decrease in levels of air pollution. This is particularly critical in view of the increased mortality, particularly an 11% increase in cardiovascular mortality as result of a 10 μgm−3 increase of PM2.5 (Bourdrel et al., 2017). Additionally, PM2.5 has been identified as the 5th risk factor of mortality, with 59% of those occurring in East and South Asia (Cohen et al., 2017). Therefore, it would be worthwhile to explore the impact of lower levels of particulate matter on the general health of the population in the DMR once the appropriate data are available for analysis. Author statement Shouraseni Sen Roy and Robert C. Balling Jr.: Conceptualization, Methodology, Data curation, Visualization, Investigation, Writing- Original draft preparation, Reviewing, and Editing. ==== Refs References Aneja V.P. Agarwal A. Roelle P.A. Phillips S.B. Tong Q.S. Watkins N. Yablonsky R. Measurements and analysis of criteria pollutants in New Delhi, India Environment International 27 2001 35 42 11488388 Balachandran S. Meena B.R. Khillare P.S. Particle size distribution and its elemental composition in the ambient air of Delhi Environment International 26 2000 49 54 11345737 Bhakta R. Khillare P.S. Jyethi D.S. Atmospheric particulate matter variations and comparison of two forecasting models for two Indian megacities Aerosol Science and Engineering 3 2 2019 54 62 Bourdrel T. Bind M.A. Béjot Y. Morel O. Argacha J.F. Cardiovascular effects of air pollution Archives of Cardiovascular Diseases 110 2017 634 642 28735838 Business Today Lockdown reduces Delhi pollution; air quality turns 'good' from 'hazardous.' Business Today https://www.businesstoday.in/latest/trends/lockdown--reduces-delhi-pollution-air-quality-turns-good-from-hazardous/story/399397.html 2020 Chowdhury S. Dey S. Cause-specific premature death from ambient PM2.5 exposure in India: Estimate adjusted for baseline mortality Environment International 91 2016 283 290 27063285 Clark P.J. Evans F.C. Distance to nearest neighbor as a measure of spatial relationships in populations Ecology 35 1954 445 453 Cohen A.J. Brauer M. Burnett R. Estimates and 25-year trends of the global burden of disease attributable to ambient air pollution: An analysis of data from the global burden of diseases study 2015 The Lancet 389 2017 1907 1918 Ellis-Petersen H. Ratcliffe R. Cowie S. Daniels J.P. Kuo L. It's positively alpine!: Disbelief in big cities as air pollution falls. The Guardian https://www.theguardian.com/environment/2020/apr/11/positively-alpine-disbelief-air-pollution-falls-lockdown-coronavirus 2020 Ghude S.D. Chate D.M. Jena C. Beig G. Kumar R. Barth M.C. Pfister G.G. Fadnavis S. Pithani P. Premature mortality in India due to PM2.5 and ozone exposure Geophysical Research Letters 43 2016 4650 4658 Gurjar B.R. Ravindra K. Nagpure A.S. Air pollution trends over India megacities and their local-to-global implications Atmospheric Environment 142 2016 475 495 Gurjar B.R. van Aardenne J.A. Lelieveld J. Mohan M. Emission estimates and trends (1990–2000) for mega city Delhi and implications Atmospheric Environment 38 2004 5663 5681 Jain M. Dawa D. Mehta R. Dimri A.P. Pandit M.K. Monitoring land use change and its drivers in Delhi, India using multi-temporal satellite data Modeling Earth Systems and Environment 2 1 2016 19 Jain S. Sharma S.K. Mandal T.K. Saxena M. Source apportionment of PM10 in Delhi, India using PCA/APCS, UNMIX and PMF Particuology 37 2018 107 118 Kandlikar M. Air pollution at a hotspot location in Delhi: Detecting trends, seasonal cycles and oscillations Atmospheric Environment 41 2007 5934 5947 Kumari P. Toshniwal D. Impact of lockdown measures during COVID-19 on air quality–A case study of India International Journal of Environmental Health Research 2020 1 8 Lelieveld J.O. Crutzen P.J. Ramanathan V. Andreae M.O. Brenninkmeijer C.A.M. Campos T. Cass G.R. Dickerson R.R. Fischer H. De Gouw J.A. Hansel A. The Indian ocean experiment: Widespread air pollution from south and southeast Asia Science 291 5506 2001 1031 1036 11161214 Liu Z. Sen Roy S. Spatial patterns of seasonal level diurnal variations of ozone and respirable suspended particulates in Hong Kong The Professional Geographer 67 1 2014 17 27 10.1080/00330124.2014.886922 Mahato S. Pal S. Ghosh K.G. Effect of lockdown amid COVID-19 pandemic on air quality of the megacity Delhi, India The Science of the Total Environment 2020 139086 Nagar P.K. Sharma M. Das D. A new method for trend analyses in PM10 and impact of crop residue burning in Delhi, Kanpur and Jaipur, India Urban Climate 27 2019 193 203 Nagpure A.S. Sharma K. Gurjar B.R. Traffic induced emission estimates and trends (2000–2005) in megacity Delhi Urban Climate 4 2013 61 73 Nelson H. Harmonic analysis 1983 Addison-Wesley Reading, Massachusetts Oliver M.A. Webster R. Kriging: A method of interpolation for geographical information systems International Journal of Geographical Information Systems 4 1990 313 332 O'Shea P.M. Sen Roy S. Singh R.B. Diurnal variations in the spatial patterns of air pollution across the Delhi metropolitan region Theoretical and Applied Climatology 124 3 2015 609 620 10.1007/s00704-015-1441-y Patel K. Airborne particle levels plummet in northern India https://earthobservatory.nasa.gov/images/146596/airborne-particle-levels-plummet-in-northern-india 2020 Ratnam M.V. Santhi Y.D. Rajeevan M. Rao S.V.B. Diurnal variability of stability indices observed using radiosonde observations over a tropical station: Comparison with microwave radiometer measurements Atmospheric Research 124 2013 21 33 Sahu S.K. Beig G. Parkhi N.S. Emissions inventory of anthropogenic PM2. 5 and PM10 in Delhi during commonwealth games 2010 Atmospheric Environment 45 34 2011 6180 6190 Saxena M. Sharma A. Sen A. Saxena P. Mandal T.K. Sharma S.K. Sharma C. Water soluble inorganic species of PM10 and PM2. 5 at an urban site of Delhi, India: Seasonal variability and sources Atmospheric Research 184 2017 112 125 Saxena M. Singh D.P. Saud T. Gadi R. Singh S. Sharma S. Mandal T.K. Study on particulate polycyclic aromatic hydrocarbons over Bay of Bengal Atmospheric Research 145– 146 2014 205 217 Sen Roy S. Impact of lunar cycle on the precipitation in India Geophysical Research Letters 33 1 2006 Sen Roy S. The impacts of ENSO, PDO, and local SSTs on winter precipitation in India Physical Geography 27 2006 464 474 Sen Roy S. Rahman A. Ahmed S. Shahfahad Ahmad I.A. Alarming groundwater depletion in the Delhi metropolitan region: A long-term assessment Environmental Monitoring and Assessment 192 2020 10.1007/s10661-020-08585-8 Sen A. Ahammed Y.N. Banerjee T. Chatterjee A. Choudhuri A.K. Das T. Deb N.C. Dhir A. Goel S. Khan A.H. Mandal T.K. Murari V. Rao P.S. Saxena M. Sharma S.K. Sharma A. Vachaspati C.V. Spatial variability in ambient atmospheric fine and coarse mode aerosols over indo-Gangetic plains, India and adjoining oceans during the onset of summer monsoons Atmos. Poll. Res. 7 2016 521 532 Sharma S.K. Mandal T.K. Jain S. Saraswati S.A. Saxena M. Source apportionment of PM2.5 in Delhi, India using PMF model Bulletin of Environmental Contamination and Toxicology 97 2016 286 293 27209541 Shehzad K. Sarfraz M. Shah S.G.M. The impact of COVID-19 as a necessary evil on air pollution in India during the lockdown Environmental Pollution 266 2020 115080 32634726 Singh V. Singh S. Biswal A. Kesarkar A.P. Mor S. Ravindra K. Diurnal and temporal changes in air pollution during COVID-19 strict lockdown over different regions of India Environmental Pollution 266 2020 115368 32829030 Singh K. Tiwari S. Jha A.K. Aggarwal S.G. Bisht D.S. Murty B.P. Khan Z.H. Gupta P.K. Mass-size distribution of PM 10 and its characterization of ionic species in fine (PM 2.5) and coarse (PM 10− 2.5) mode, New Delhi, India Natural Hazards 68 2 2013 775 789 Sudheer A.K. Sarin M.M. Carbonaceous aerosols in MABL of Bay of bengal: Influence of continental outflow Atmospheric Environment 42 18 2008 4089 4100 Tiwari S. Hopke P.K. Pipal A.S. Srivastava A.K. Bisht D.S. Tiwari S. Singh A.K. Soni V.K. Attri S.D. Intra-urban variability of particulate matter (PM2. 5 and PM10) and its relationship with optical properties of aerosols over Delhi, India Atmospheric Research 166 2015 223 232 Venter Z.S. Aunan K. Chowdhury S. Lelieveld J. COVID-19 lockdowns cause global air pollution declines with implications for public health risk https://www.medrxiv.org/content/10.1101/2020.04.10.20060673v1.article-metrics 2020 Wu X. Nethery R.C. Sabath B.M. Braun D. Dominici F. Exposure to air pollution and COVID-19 mortality in the United States 2020 medRxiv 2020

0

PMC9753958

NO-CC CODE

2022-12-16 23:26:23

no

Prog Cardiovasc Dis. 2021 Aug 17 July-August; 67:A1-A3

latin-1

Prog Cardiovasc Dis

10.1016/S0033-0620(21)00083-9

oa_other

==== Front JACC Asia JACC Asia JACC Asia 2772-3747 The Authors. Published by Elsevier on behalf of the American College of Cardiology Foundation. S2772-3747(22)00235-6 10.1016/j.jacasi.2022.09.005 Original Research Real-World Management of Pharmacological Thromboprophylactic Strategies for COVID-19 Patients in Japan From the CLOT-COVID Study Hayashi Hiroya MD a∗ Izumiya Yasuhiro MD a Fukuda Daiju MD a Wakita Fumiaki MD b Mizobata Yasumitsu MD b Fujii Hiromichi MD c Yachi Sen MD d Takeyama Makoto MD d Nishimoto Yuji MD e Tsujino Ichizo MD f Nakamura Junichi MD f Yamamoto Naoto MD g Nakata Hiroko MD h Ikeda Satoshi MD i Umetsu Michihisa MD j Aikawa Shizu MD k Satokawa Hirono MD l Okuno Yoshinori MD m Iwata Eriko MD n Ogihara Yoshito MD o Ikeda Nobutaka MD p Kondo Akane MD q Iwai Takehisa MD r Yamada Norikazu MD s Ogawa Tomohiro MD t Kobayashi Takao MD g Mo Makoto MD u Yamashita Yugo MD m on behalf of the CLOT-COVID Study Investigators a Department of Cardiovascular Medicine, Osaka Metropolitan University Graduate School of Medicine, Osaka, Japan b Department of Traumatology and Critical Care Medicine, Osaka Metropolitan University Graduate School of Medicine, Osaka, Japan c Department of Intensive Care Medicine, Osaka Metropolitan University Graduate School of Medicine, Osaka, Japan d Japan Community Health Care Organization Tokyo Shinjuku Medical Center, Tokyo, Japan e Hyogo Prefectural Amagasaki General Medical Center, Amagasaki, Japan f Hokkaido University Hospital, Sapporo, Japan g Hamamatsu Medical Center, Hamamatsu, Japan h Yokosuka General Hospital Uwamachi, Yokosuka, Japan i Nagasaki University Graduate School of Biomedical Sciences, Nagasaki, Japan j Tohoku University Hospital, Sendai, Japan k Tsukuba Medical Center Hospital, Tsukuba, Japan l Fukushima Red Cross Hospital, Fukushima, Japan m Kyoto University Hospital, Kyoto, Japan n Nankai Medical Center Japan Community Health Care Organization, Saiki, Japan o Mie University Hospital, Tsu, Japan p Toho University Ohashi Medical Center, Tokyo, Japan q Shikoku Medical Center for Children and Adults, Zentsuji, Japan r Tsukuba Vascular Center, Ibaraki, Japan s Kuwana City Medical Center, Kuwana, Japan t Fukushima Daiich Hospital, Fukushima, Japan u Yokohama Minami Kyosai Hospital, Yokohama, Japan ∗ Address for correspondence: Dr Hiroya Hayashi, Department of Cardiovascular Medicine, Osaka Metropolitan University Graduate School of Medicine, 1-4-3 Asahimachi, Abenoku, Osaka 545-8585, Japan. 15 12 2022 12 2022 15 12 2022 2 7 897907 23 5 2022 6 9 2022 6 9 2022 © 2022 The Authors 2022 Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. Background Data on prophylactic anticoagulation are important in understanding the current issues, unmet needs, and optimal management of Japanese COVID-19 patients. Objectives This study aimed to investigate the clinical management strategies for prophylactic anticoagulation of COVID-19 patients in Japan. Methods The CLOT-COVID study was a multicenter observational study that enrolled 2,894 consecutive hospitalized patients with COVID-19. The study population consisted of 2,889 patients (after excluding 5 patients with missing data); it was divided into 2 groups: patients with pharmacological thromboprophylaxis (n = 1,240) and those without (n = 1,649). Furthermore, we evaluated the 1,233 patients who received prophylactic anticoagulation—excluding 7 patients who could not be classified based on the intensity of their anticoagulants—who were then divided into 2 groups: patients receiving prophylactic anticoagulant doses (n = 889) and therapeutic anticoagulant doses (n = 344). Results The most common pharmacological thromboprophylaxis anticoagulant was unfractionated heparin (68.2%). The severity of COVID-19 at admission was a predictor of the implementation of pharmacological thromboprophylaxis in the multivariable analysis (moderate vs mild: OR: 16.6; 95% CI:13.2-21.0; P < 0.001, severe vs mild: OR: 342.6, 95% CI: 107.7–1090.2; P < 0.001). It was also a predictor of the usage of anticoagulants of therapeutic doses in the multivariable analysis (moderate vs mild: OR: 2.10; 95% CI: 1.46-3.02; P < 0.001, severe vs mild: OR: 5.96; 95% CI: 3.91-9.09; P < 0.001). Conclusions In the current real-world Japanese registry, pharmacological thromboprophylaxis, especially anticoagulants at therapeutic doses, was selectively implemented in COVID-19 patients with comorbidities and severe COVID-19 status at admission. Central Illustration Key Words COVID-19 anticoagulation COVID-19-associated coagulopathy therapeutic anticoagulants thromboprophylaxis venous thromboembolism Abbreviations and Acronyms DOAC, direct oral anticoagulants LMWH, low-molecular-weight heparin UFH, unfractionated heparin VTE, venous thromboembolism ==== Body pmcThe COVID-19 pandemic, caused by SARS-CoV-2, has infected millions of people worldwide and become a major health problem.1, 2, 3 COVID-19 has also been associated with several cardiovascular complications, including venous and arterial thrombosis.4, 5, 6, 7 In particular, COVID-19–associated coagulopathy has been reported to cause the formation of thrombi in the capillary-alveolar interface of large and small blood vessels in the lungs, contributing to worsening respiratory failure and highlighting the importance of anticoagulation therapy.8 , 9 Because prophylactic anticoagulation is thought to be effective in preventing thrombosis and disease complications, several current international guidelines recommend prophylactic anticoagulation for all COVID-19 inpatients.10, 11, 12 However, the risk of thrombosis and hemorrhagic adverse events associated with prophylactic anticoagulation in patients with COVID-19 can vary widely according to patient characteristics and ethnic differences. Historically, routine prophylactic anticoagulation is not recommended for acutely ill hospitalized patients in Japan, unlike in other countries. In fact, the latest Japanese domestic COVID-19 Clinical Practice Guidelines by the Ministry of Health, Labor, and Welfare in Japan have not recommended routine prophylactic anticoagulation for all inpatients with COVID-19; this is because of the potentially lower risk of thrombosis and higher risk of bleeding in Japanese patients,13 , 14 and this fact could lead to the requirement of widely varying management strategies—including the type, intensity, and duration of prophylactic anticoagulation. Data on the current real-world management strategies of prophylactic anticoagulation are important for understanding the issues, unmet needs, and optimal management of patients with COVID-19. However, data available on these issues are scarce in Japan. Therefore, the purpose of the current study was to investigate the current real-world management strategies for prophylactic anticoagulation by using a large-scale multicenter observational database of patients with COVID-19 in Japan. Methods Study population The CLOT-COVID Study (thrombosis and antiCoaguLatiOn Therapy in patients with COVID-19 in Japan Study; UMIN000045800) is a physician-initiated, retrospective, multicenter cohort study enrolling consecutive hospitalized patients with COVID-19 from 16 centers in Japan from April 2021 to September 2021. The design of this study has been previously reported in detail.15 We collected the data of 2,894 consecutive patients who were diagnosed with COVID-19 with a positive polymerase chain reaction test. Results were retrieved from hospital databases. After excluding 5 patients with incomplete detailed data on anticoagulants, the current study population consisted of 2,889 patients, who were divided into 2 groups: patients with pharmacological thromboprophylaxis and those without (Figure 1 ). Furthermore, for the comparison between anticoagulants of prophylactic doses and those of therapeutic doses, we evaluated 1,233 patients with prophylactic anticoagulation (after excluding 7 patients with anticoagulants that could not be classified based on the intensity of anticoagulants, including argatroban), who were subsequently divided into 2 groups: patients receiving prophylactic anticoagulant doses and those receiving anticoagulants at therapeutic doses.Figure 1 Study Flowchart After excluding 5 patients with incomplete detailed data on anticoagulants, the current study population consisted of 2,889 patients who were divided into 2 groups: patients with pharmacological thromboprophylaxis and those without. Furthermore, we evaluated 1,233 patients with prophylactic anticoagulation who were subsequently divided into 2 groups: patients receiving prophylactic anticoagulant doses and those receiving anticoagulants at therapeutic doses. Pharmacological thromboprophylactic management was evaluated by the anticoagulants administrated during hospitalization (excluding their use in thrombosis treatment), which were divided into the following 7 groups according to the types and doses of the anticoagulants: unfractionated heparin (UFH) of a prophylactic dose, UFH of a therapeutic dose, low-molecular-weight heparin (LMWH) of a prophylactic dose, LMWH of a therapeutic dose, direct oral anticoagulants (DOACs), warfarin, and others. Therapeutic dose UFH was defined as the administration of UFH targeting a therapeutic range, referencing the activated partial thromboplastin time. UFH of the prophylactic dose was defined as the administration of UFH at a fixed dose without referencing the activated partial thromboplastin time. The prophylactic anticoagulant doses included UFH of a prophylactic dose and LMWH of a prophylactic dose. Therapeutic doses of anticoagulants included UFH of a therapeutic dose, LMWH of a therapeutic dose, DOACs, and warfarin. LMWH for acutely ill patients as pharmacological thromboprophylaxis and treatment for acute thrombosis is not approved by Japanese insurance, and we had a prophylactic dose for LMWH in the package insert. Thus, anticoagulation therapy in the current study was evaluated based on either standard dose or therapeutic dose. Ethics approval and consent to participate All procedures were in accordance with the Declaration of Helsinki. The relevant review boards and ethics committees of all participating centers approved the research protocol. Written, informed consent from patients was waived because we used clinical information obtained in routine clinical practice. This method is concordant with the guidelines for epidemiological studies issued by the Ministry of Health, Labor, and Welfare in Japan. Data collection Patient data and follow-up information were collected using an electronic reporting form. Data on patient characteristics, pharmacological thromboprophylaxis management, and clinical outcomes were collected from hospital charts or hospital databases according to prespecified definitions. The physicians at each institution were responsible for data entry into the electronic case report form. In addition, data were manually checked for missing or contradictory inputs and values outside the expected ranges at the research-based office. Definitions for patient characteristics Hypertension was diagnosed if the peripheral blood pressure was >140/90 mm Hg or if the patient was taking medication for hypertension. The presence of diabetes was diagnosed using hemoglobin A1c (National Glycohemoglobin Standardization Program, 6.5%) as the standard or was assumed if the patient was taking medication for the treatment of diabetes. Heart disease was defined as heart disorders such as heart failure, arrhythmias, angina pectoris, and a history of myocardial infarction. Heart failure was diagnosed if the patient had a history of hospitalization for heart failure, if the patient displayed symptoms caused by heart failure (New York Heart Association [NYHA] functional class ≥2), or if the left ventricular ejection fraction was ≤40%. Respiratory disease was defined as the presence of a chronic lung disorder such as asthma, chronic obstructive pulmonary disease, and restrictive lung disease. Patients with active cancer were defined as those receiving treatment for cancer (such as chemotherapy or radiotherapy), those scheduled to undergo cancer surgery with metastasis to other organs, and/or those with terminal cancer.16 A history of major bleeding was diagnosed if the patient had a major bleeding event (classified by the International Society of Thrombosis and Hemostasis), which consisted of a reduction in the hemoglobin level by at least 2 g/dL, transfusion of at least 2 U of blood, or symptomatic bleeding in a critical area or organ.17 The severity of COVID-19 was classified as mild, moderate, or severe.13 Patients with mild COVID-19 were defined as those who did not require oxygen, patients with moderate COVID-19 were defined as those who required oxygen, and patients with severe COVID-19 were defined as those who required mechanical ventilation or extracorporeal membrane oxygenation. The patients with the worst severity of COVID-19 during hospitalization were classified into 3 groups: mild, moderate, and severe status—or death at discharge.13 Clinical outcomes The outcomes measured in the current study were thrombosis, major bleeding, and all-cause mortality during hospitalization. Thrombosis includes venous thromboembolism (VTE), ischemic stroke, myocardial infarction, and systemic arterial thromboembolism. VTE was defined as pulmonary embolism and/or deep vein thrombosis objectively confirmed by imaging examinations (ultrasound, contrast-enhanced computed tomography, ventilation-perfusion lung scintigraphy, pulmonary angiography, or contrast venography) or an autopsy. Ischemic stroke was defined as a stroke requiring or prolonging hospitalization with symptoms lasting more than 24 hours. Myocardial infarction was defined according to universal myocardial infarction guidelines.18 Major bleeding was diagnosed as International Society of Thrombosis and Hemostasis major bleeding, which consisted of a reduction in the hemoglobin level by at least 2 g/dL, transfusion of at least 2 U of blood, or symptomatic bleeding in a critical area or organ.17 Statistical analysis Categorical variables are presented as numbers and percentages. Continuous variables are presented as the mean ± SD or median (IQR) based on their distributions. Categorical variables were compared using the chi-square test when appropriate; otherwise, Fisher exact test was used. Continuous variables were compared using Student's t-test or Wilcoxon’s rank-sum test based on their distributions. Clinical outcomes were presented as the number of events and percentages. We also conducted a stratified analysis by severity of COVID-19. In the current study, to investigate the clinical characteristics that were associated with the implementation of pharmacological thromboprophylaxis and usage of anticoagulants at therapeutic doses, we estimated the adjusted ORs and their 95% CIs using a multivariable logistic regression model. We calculated the crude ORs by univariate analysis for all patient characteristics: including baseline characteristics, comorbidities, the severity of COVID-19 at admission, and the worsening of COVID-19 severity during hospitalization. We selected potential variables for multivariate analysis based on variables with P values <0.05 by univariate analysis. Furthermore, we also conducted the multivariable logistic regression analysis as an exploratory analysis to investigate the potential impact of anticoagulation strategies on clinical events considering several confounding factors. Statistical significance was set at P < 0.05. All analyses were performed using JMP version 13.0.0 (SAS Institute). Results Characteristics and clinical outcomes in patients with and without pharmacological thromboprophylaxis Patient characteristics and a comparison of patients with and without pharmacological thromboprophylaxis are shown in Table 1 . Of the 2,889 patients, 1,240 patients (42.9%) received pharmacological thromboprophylaxis. Patients with pharmacological thromboprophylaxis were more often men (71.8% vs 60.1%; P < 0.001), were older (age 58.2 years vs 48.5 years; P < 0.001), had a higher prevalence of body mass index >30 kg/m2 (20.1% vs 12.7%; P < 0.001), and had a higher median D-dimer level at admission (1.1 μg/mL vs 0.6 μg/mL; P < 0.001) than those without. Regarding comorbidities, hypertension (40.6% vs 22.2%; P < 0.001), diabetes mellitus (26.8% vs 16.0%; P < 0.001), and heart disease (12.3% vs 6.2%; P < 0.001) were more common in patients with pharmacological thromboprophylaxis. Patients with pharmacological thromboprophylaxis also showed more severe status of COVID-19 at admission (mild: 26.1% vs 85.6%; moderate: 55.7% vs 14.2%; and severe: 18.2% vs 0.2%; P < 0.001) and a higher prevalence of worsening COVID-19 severity during hospitalization (33.6% vs 17.0%; P < 0.001). Patient characteristics and clinical outcomes stratified by severity of COVID-19 are shown in Supplemental Table 1.Table 1 Patient Characteristics and Clinical Outcomes Comparing Patients With and Without Pharmacological Thromboprophylaxis (N = 2,889) Pharmacological Thromboprophylaxis (+) (n = 1,240) Pharmacological Thromboprophylaxis (−) (n = 1,649) P Value Baseline characteristics  Age, y 58.2 ± 14.5 48.5 ± 19.0 <0.001  Men 890 (71.8) 991 (60.1) <0.001  Body mass index, kg/m2 26.3 ± 5.6 24.5 ± 5.2 <0.001  Body mass index >30 kg/m2 249 (20.1) 209 (12.7) <0.001  D-dimer level at admission, μg/mL 1.1 (0.7-1.8) 0.6 (0.5-0.9) <0.001 Comorbidities  Hypertension 503 (40.6) 366 (22.2) <0.001  Diabetes mellitus 332 (26.8) 264 (16.0) <0.001  Heart disease 153 (12.3) 102 (6.2) <0.001  Respiratory Disease 135 (10.9) 163 (9.9) 0.38  Active cancer 33 (2.7) 27 (1.6) 0.06  History of major bleeding 15 (1.2) 12 (0.7) 0.18  History of VTE 10 (0.8) 5 (0.3) 0.06  Duration of hospitalization, days 13 (9-20) 7 (5-10) <0.001 Severity of COVID-19 at admission  Mild 323 (26.1) 1,412 (85.6) <0.001  Moderate 691 (55.7) 234 (14.2)  Severe 226 (18.2) 3 (0.2) Worst severity of COVID-19 during hospitalization  Mild 124 (10.0) 1,158 (70.2) <0.001  Moderate 746 (60.2) 482 (29.2)  Severe or death at discharge 370 (29.8) 10 (0.6)  Worsening severity of COVID-19 during hospitalization 417 (33.6) 280 (17.0) <0.001 Detailed anticoagulants of pharmacological thromboprophylaxis  Anticoagulation therapy with a prophylactic dose 889 (71.2)  Unfractionated heparin of a prophylactic dose 685 (55.2) - -  Low-molecular-weight heparin of a prophylactic dose 204 (16.5) - -  Anticoagulation therapy with a therapeutic dose 344 (27.7)  Unfractionated heparin of a therapeutic dose 161 (13.0) — —  Low-molecular-weight heparin of a therapeutic dose 0 (0) — —  Warfarin 19 (1.5) — —  DOACs 164 (13.2) — —  Others 7 (0.6) Clinical outcomea  Thrombosis 52 (4.2 [3.2-5.5]) 2 (0.1 [0.0-0.5]) <0.001  VTE 37 (3.0 [2.2-4.1]) 1 (0.06 [0.0-0.4]) <0.001  Bleeding 51 (4.1 [3.1-5.4]) 6 (0.4 [0.1-0.8]) <0.001  All-cause death 130 (10.5 [8.9-12.3]) 28 (1.7 [1.2-2.5]) <0.001 Values are mean ± SD, n (%), or median (IQR), unless otherwise indicated. DOAC = direct oral anticoagulant; VTE = venous thromboembolism. a The clinical outcomes are presented as numbers of events and percentages with the 95% CIs Event rates were 4.2% (95% CI: 3.2%-5.5%) for thrombosis, 3.0% (95% CI: 2.2%-4.1%) for VTE, 4.1% (95% CI: 3.1%-5.4%) for major bleeding, and 10.5% for all-cause death (95% CI: 8.9%-12.3%) in the patients with pharmacological thromboprophylaxis. In the patients without pharmacological thromboprophylaxis, the rates were 0.1% (95% CI: 0.0%-0.5%) for thrombus, 0.06% (95% CI: 0.0%-0.4%) for VTE, 0.4% (95% CI: 0.1%-0.8%) for major bleeding, and 1.7% (95% CI: 1.2%-2.5%) for all-cause death (Table 1). In multivariate analysis, severe COVID-19 at admission was significantly associated with all clinical events (Supplemental Table 2). Patient characteristics associated with the implementation of pharmacological thromboprophylaxis The multivariable logistic regression model revealed that older age (OR: 1.02; 95% CI: 1.01-1.02; P < 0.001), men (OR: 1.57; 95% CI: 1.26-1.95; P < 0.001), body mass index >30 kg/m2 (OR: 1.91; 95% CI: 1.44-2.53; P < 0.001), high D-dimer levels at admission (OR: 1.06; 95% CI: 1.02-1.09; P = 0.003), severity of COVID-19 at admission (moderate vs mild: OR: 16.6; 95% CI: 13.2-21.0; P < 0.001, severe vs mild: OR: 342.6; 95% CI: 107.7-1,090.2; P < 0.001), and worsening of COVID-19 severity during hospitalization (OR: 4.83; 95% CI: 3.77-6.18; P < 0.001) were predictors of the implementation of pharmacological thromboprophylaxis during hospitalization in multivariable analysis (Table 2 ).Table 2 Patient Characteristics Associated With the Implementation of Pharmacological Thromboprophylaxis During Hospitalization Univariate Analysis Multivariate Analysis Crude OR (95% CI) P Value Adjusted OR (95% CI) P Value Age (per 1 y) 1.03 (1.03-1.04) <0.001 1.02 (1.01-1.02) <0.001 Men (vs women) 1.69 (1.44-1.98) <0.001 1.57 (1.26-1.95) <0.001 Body mass index >30 kg/m2 1.73 (1.42-2.12) <0.001 1.91 (1.44-2.53) <0.001 D-dimer level at admission (per 1 μg/mL) 1.26 (1.19-1.34) <0.001 1.06 (1.02-1.09) 0.003 Hypertension 2.40 (2.04-2.82) <0.001 1.17 (0.92-1.49) 0.22 Diabetes mellitus 1.92 (1.60-2.30) <0.001 0.95 (0.74-1.23) 0.69 Heart disease 2.10 (1.61-2.72) <0.001 1.24 (0.86-1.81) 0.25 Respiratory disease 1.11 (0.88-1.42) 0.38 — — Active cancer 1.64 (0.98-2.75) 0.06 — — History of major bleeding 1.67 (0.78-3.58) 0.19 — — History of VTE 2.67 (0.91-7.84) 0.07 — — Severity of COVID-19 at admission  Moderate (vs mild) 12.9 (10.7-15.6) <0.001 16.6 (13.2-21.0) <0.001  Severe (vs mild) 329.3 (104.7-1,035.4) <0.001 342.6 (107.7-1,090.2) <0.001  Worsening severity of COVID-19 during hospitalization 2.48 (2.08-2.95) <0.001 4.83 (3.77-6.18) <0.001 VTE = venous thromboembolism. Characteristics and clinical outcomes in patients with anticoagulants of prophylactic and therapeutic doses Of 1,233 patients, 889 (72.1%) and 344 (27.9%) received prophylactic-dose anticoagulants and therapeutic-dose anticoagulants, respectively (Figure 2 ). In particular, 77.1% and 22.9% of prophylactic-dose anticoagulants consisted of UFH and LMWH, respectively, while 46.8%, 5.5%, and 47.7% of therapeutic-dose anticoagulants were UFH, warfarin, and DOACs, respectively. Table 3 shows that patients with therapeutic-dose anticoagulants were older (56.8 years vs 62.2 years; P < 0.001), had higher median D-dimer levels at admission (1.1 μg/mL vs 1.3 μg/mL; P < 0.001), a higher prevalence of hypertension (38.1% vs 47.7%; P = 0.002), more heart disease (7.4% vs 24.7%; P < 0.001), and increased VTE history (0.2% vs 2.3%; P < 0.001) than those with prophylactic-dose anticoagulants. Patients who were administered anticoagulants at therapeutic doses showed a more severe COVID-19 status at admission than those with anticoagulants of prophylactic doses (mild: 30.1% vs 15.1%; moderate: 56.9% vs 52.9%; and severe: 12.9% vs 32.0%; P < 0.001), whereas there was no significant difference in the prevalence of worsening COVID-19 severity during hospitalization between the 2 groups (32.7% vs 36.6%; P = 0.20).Figure 2 Detailed Anticoagulants of Pharmacological Thromboprophylaxis Of 1,233 patients, 889 patients (72.1%) and 344 patients (27.9%) received prophylactic-dose anticoagulants and therapeutic-dose anticoagulants, respectively. In particular, 77.1% and 22.9% of prophylactic-dose anticoagulants consisted of unfractionated heparin (UFH) and low-molecular-weight heparin (LMWH), respectively, whereas 46.8%, 5.5%, and 47.7% of therapeutic-dose anticoagulants were UFH, warfarin, and direct oral anticoagulants (DOACs), respectively. Table 3 Patient Characteristics and Clinical Outcomes Comparing Patients With Anticoagulants of Prophylactic or Therapeutic Doses (N = 1,233) Anticoagulants of Prophylactic Doses (n = 889) Anticoagulants of Therapeutic Doses (n = 344) P Value Baseline characteristics  Age, y 56.8 ± 14.2 62.2 ± 14.3 <0.001  Men 643 (72.3) 243 (70.6) 0.56  Body mass index, kg/m2 26.3 ± 5.3 26.1 ± 6.0 0.69  Body mass index >30 kg/m2 172 (19.4) 75 (21.8) 0.33  D-dimer level at admission, μg/mL 1.1 (0.7-1.6) 1.3 (0.8-2.8) <0.001 Comorbidities  Hypertension 339 (38.1) 164 (47.7) 0.002  Diabetes mellitus 245 (27.6) 86 (25.0) 0.36  Heart disease 66 (7.4) 85 (24.7) <0.001  Respiratory disease 94 (10.6) 41 (11.9) 0.50  Active cancer 23 (2.6) 10 (2.9) 0.76  History of major bleeding 8 (0.9) 6 (1.7) 0.21  History of VTE 2 (0.2) 8 (2.3) <0.001  Duration of hospitalization 13 (9-19) 15 (10-23) <0.001 Severity of COVID-19 at admission  Mild 268 (30.1) 52 (15.1) <0.001  Moderate 506 (56.9) 182 (52.9)  Severe 115 (12.9) 110 (32.0) Worst severity of COVID-19 during hospitalization  Mild 92 (10.3) 29 (8.4) <0.001  Moderate 596 (67.0) 147 (42.7)  Severe or death at discharge 201 (22.6) 168 (48.9)  Worsening severity of COVID-19 during hospitalization 291 (32.7) 126 (36.6) 0.20 Clinical outcomea  Thrombosis (event rate [95%CI]) 31 (3.5 [2.5-4.9]) 2 (0.6 [0.0-2.2]) 0.04  VTE (event rate [95%CI]) 21 (2.4 [1.5-3.6]) 16 (4.7 [2.8-7.5]) 0.04  Bleeding (event rate [95%CI]) 20 (2.3 [1.4-3.5]) 3 (0.9 [0.2-2.7]) <0.001  All-cause death (event rate [95%CI]) 60 (6.8 [5.3-8.6]) 70 (20.3 [16.4-24.9]) <0.001 Values are mean ± SD, n (%), or median (IQR), unless otherwise indicated. VTE = venous thromboembolism. a The clinical outcomes are presented as numbers of events and percentages with the 95% CIs Event rates were 3.5% (95% CI: 2.5%-4.9%) for thrombosis, 2.4% (95% CI: 1.5%-3.6%) for VTE, 2.3% (95% CI: 1.4%-3.5%) for major bleeding, and 6.8% (95% CI: 5.3%-8.6%) for all-cause death in the patients with anticoagulants of prophylactic doses. In the patients with anticoagulants of therapeutic doses, the rates were 0.6% (95% CI: 0.0%-2.2%) for thrombus, 4.7% (95% CI: 2.8%-7.5%) for VTE, 0.9% (95% CI: 0.2%-2.7%) for major bleeding, and 20.3% (95% CI: 16.4%-24.9%) for all-cause death (Table 3). In multivariate analysis, severe COVID-19 at admission was significantly associated with all clinical events (Supplemental Table 3). Patient characteristics associated with the usage of therapeutic-dose anticoagulants The multivariable logistic regression model revealed that older age (OR: 1.02; 95% CI: 1.01-1.03; P < 0.001), heart disease (OR: 3.84; 95% CI: 2.60-5.68; P < 0.001), history of VTE (OR: 10.6; 95% CI: 1.93-57.8; P = 0.007), and severity of COVID-19 at admission (moderate vs mild: OR: 2.10; 95% CI: 1.46-3.02; P < 0.001, severe vs mild: OR: 5.96; 95% CI: 3.91-9.09; P < 0.001) were predictors of the therapeutic-dose anticoagulants during hospitalization in multivariable analysis (Table 4 ).Table 4 Patient Characteristics Associated With Usage of Anticoagulants of Therapeutic Doses During Hospitalization Univariate Analysis Multivariate Analysis Crude OR (95% CI) P Value Adjusted OR (95% CI) P Value Age (per 1 y) 1.03 (1.02-1.04) <0.001 1.02 (1.01-1.03) <0.001 Men (vs women) 0.92 (0.70-1.21) 0.55 — — Body mass index >30 kg/m2 1.16 (0.86-1.58) 0.33 — — D-dimer level at admission (per 1 μg/mL) 1.01 (0.99-1.01) 0.06 — — Hypertension 1.48 (1.15-1.90) 0.002 0.97 (0.73-1.30) 0.85 Diabetes mellitus 0.88 (0.66-1.17) 0.36 — — Heart disease 4.09 (2.88-5.81) <0.001 3.84 (2.60-5.68) <0.001 Respiratory disease 1.14 (0.77-1.69) 0.50 — — Active cancer 1.13 (0.53-2.39) 0.76 — — History of major bleeding 1.95 (0.67-5.68) 0.22 — — History of VTE 10.6 (2.23-50.0) 0.003 10.6 (1.93-57.8) 0.007 Severity of COVID-19 at admission  Moderate (vs mild) 1.85 (1.32-2.61) <0.001 2.10 (1.46-3.02) <0.001  Severe (vs mild) 4.93 (3.32-7.32) <0.001 5.96 (3.91-9.09) <0.001  Worsening severity of COVID-19 during hospitalization 1.19 (0.92-1.54) 0.20 — — VTE = venous thromboembolism. Discussion The main findings of the current study were as follows: 1) pharmacological thromboprophylaxis was more often administered in patients with more comorbidities and a more severe COVID-19 status at admission, as well as in patients with worsening COVID-19 severity during hospitalization; 2) the most common pharmacological thromboprophylaxis anticoagulant was UFH in Japan; and 3) therapeutic-dose anticoagulants were more often used in patients with more comorbidities and a more severe COVID-19 status at admission, although there was no significant difference in the prevalence of worsening COVID-19 severity during hospitalization between patients on anticoagulants at therapeutic doses and those receiving prophylactic anticoagulants (Central Illustration ).Central Illustration Pharmacological Thromboprophylactic Strategies for COVID-19 Patients in Japan The CLOT-COVID study was a physician-initiated, retrospective, multicenter cohort study enrolling 2,894 consecutive hospitalized patients with COVID-19 from 16 centers in Japan from April 2021 to September 2021. The current study showed the detailed types and doses of anticoagulants in real-world clinical practice and patient characteristics associated with the implementation of pharmacological thromboprophylaxis and usage of anticoagulants at therapeutic doses. DOAC = direct oral anticoagulant; LMWH = low-molecular-weight heparin; UFH = unfractionated heparin. Anticoagulation therapy for COVID-19 patients might be important because it has been reported that COVID-19 could be associated with a higher incidence of thrombotic complications than other respiratory infections, and thrombosis has been shown to lead to worse outcomes in these patients.9 , 19 It has been reported that the risk of thrombosis in COVID-19 patients varies depending on patient characteristics (including the severity of COVID-19); however, the optimal prophylactic anticoagulation strategy for COVID-19 patients remains controversial.20 , 21 Recommendations for prophylactic anticoagulation could vary by region and country,22 possibly because the threshold for hospitalization varies with the availability of health care systems, resources, and timing of the study. Furthermore, ethnic differences could be an important issue when considering the optimal anticoagulation strategies. Actually, previous studies reported that East Asian patients could reduce anti-ischemic benefits and increase bleeding risk during antithrombotic therapies compared with Caucasian patients.23 , 24 The latest Japanese domestic COVID-19 Practice Guidelines from the Ministry of Health, Labor, and Welfare have recommended that prophylactic anticoagulation should be considered for clinically unstable hospitalized patients with COVID-19 who require oxygen administration, although it has not recommended routine prophylactic anticoagulation for all hospitalized patients with COVID-19. In Japan, even mild COVID-19 patients sometimes were hospitalized to prevent the spread and severity of COVID-19 infection, and in these cases, physicians could think that patients might be at low risk of thrombosis, resulting in no pharmacological thromboprophylaxis. Thus, the management strategies for prophylactic anticoagulation in real-world clinical practice might be at the discretion of each individual physician. In the current study, 26.1% of patients with pharmacological thromboprophylaxis had mild COVID-19 at admission, while 14.4% of patients without pharmacological thromboprophylaxis had moderate and severe COVID-19 at admission, which suggested a wide variety of management strategies depending on the institution and physician. The current study also showed that pharmacological thromboprophylaxis was more often administered to patients with more comorbidities, who might have been considered at a higher risk of worsening COVID-19 severity during hospitalization—even those with mild COVID-19 severity at admission. Worsening COVID-19 severity during hospitalization was a predictor of the implementation of pharmacological thromboprophylaxis during hospitalization, which was thought to be an important characteristic for physicians in deciding management strategies. The current study showed that a variety of anticoagulants are used in daily clinical practice. Approximately one-half of the anticoagulants used for pharmacological thromboprophylaxis were prophylactic-dose UFH, followed by therapeutic-dose UFH, indicating that UFH has been commonly used in Japan. Although LMWH for acute-ill patients as pharmacological thromboprophylaxis and treatment is not approved by Japanese insurance, the current study revealed that LMWH was sometimes used prophylactically. This was because of a shortage of UFH following a rapid increase in its use, and therefore LMWH and other drugs were used with the consent of the patient. Considering the current status of pharmacological thromboprophylaxis for COVID-19, further studies are warranted to clarify the optimal usage of LMWH in Japanese patients, including the subtypes and intensity of LMWH. The current study showed that therapeutic doses of anticoagulants were more often used in older patients with more comorbidities, especially heart disease. The association between COVID-19 and atrial fibrillation (AF) has been well reported,25, 26, 27 COVID-19 patients with AF have also been characterized by old age,28 which might be partly associated with the clinical features of patients receiving anticoagulants at therapeutic doses. Although no association has been reported between preadmission anticoagulation therapy and clinical outcomes,29 these clinical features could exert some influence on clinical outcomes. The current study also showed that therapeutic-dose anticoagulants were used more often in patients with a more severe COVID-19 status at admission. However, a previous study comparing anticoagulants at therapeutic doses and anticoagulants at prophylactic doses did not show a potential benefit of anticoagulants at therapeutic doses,30 which should be a caution for Japanese clinicians. Based on these results, it is necessary to consider the choice of pharmacological thromboprophylaxis based on the risk of severe disease rather than the current choice of inpatient or outpatient. In our data, more events occurred among patients with pharmacological thromboprophylaxis, especially in those with anticoagulants of therapeutic doses. The main reasons for this could be selection bias and confounding. In fact, the effect of pharmacological thromboprophylaxis on the risk of thrombosis was reduced after adjustment for certain confounding factors. Furthermore, several other factors that were not measured in the current study could also be responsible for the observed residual effect. Previous reports have shown that the incidence of thrombosis in Japanese COVID-19 patients ranges from 1.8% to 5.3%,31, 32, 33 whereas a meta-analysis from overseas reported a rate of 17%. The incidence of major bleeding is reported to be 2.2% to 2.9% in Japan,29 , 32 and overseas meta-analyses report a rate of 3.9%.32 Of these patients, 4.4% had major bleeding without pharmacological thromboprophylaxis, and 21.4% had major bleeding with therapeutic doses of anticoagulants.34 Furthermore, the D-dimer level at admission in COVID-19 patients was reported as 0.7 to 1.9 μg/mL in Japan,13 , 31 , 32 whereas it was reported as 2.1 to 3.8 μg/mL in Western countries.35, 36, 37, 38 These results could suggest a different risk of clinical adverse events depending on different racial groups, which would suggest the necessity of pharmacological thromboprophylaxis specifically for the Japanese population. Study limitations First, this was an observational study, which can be subject to various biases inherent in the observational study design. Therapeutic decision-making—including pharmacological thromboprophylaxis—was left to the sole discretion of the attending physicians. Second, the detailed timing and duration of pharmacological thromboprophylaxis were not evaluated in this study. Third, this study does not have an observation period long enough to understand trends, and we did not investigate the exact timing of clinical events in the current study. Finally, the current study evaluated Japanese patients and discussed some domestic medical issues in Japan; therefore, the results should be carefully extrapolated to patients from different regions and countries. Conclusions In the current real-world Japanese registry, pharmacological thromboprophylaxis and anticoagulants at therapeutic doses were implemented on COVID-19 patients with comorbidities and severe COVID-19 status at admission. The implementation of pharmacological thromboprophylaxis was also associated with worsening COVID-19 severity during hospitalization.Perspectives COMPETENCY IN MEDICAL KNOWLEDGE: The current study showed that pharmacological thromboprophylaxis was more often performed in patients with more comorbidities who might be considered at a higher risk of worsening severity of COVID-19 during hospitalization, even with mild severity of COVID-19 at admission. TRANSLATIONAL OUTLOOK: It is necessary to consider the choice of pharmacological thromboprophylaxis based on the risk of disease severity rather than the current choice of inpatient or outpatient. Funding Support and Author Disclosures The CLOT-COVID study is partially supported by research funding from Fujiwara Memorial Foundation and research funding from the Foundation Kyoto Health Care Society. The research funding had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; and preparation, review, or approval of the manuscript. The authors have reported that they have no relationships relevant to the contents of this paper to disclose. Appendix Supplemental Tables 1–3 Acknowledgements The authors appreciate the support and collaboration of the Japanese Society of Phlebology and the Japanese Society of Pulmonary Embolism Research through the current study. The authors are indebted to Ms Emi Kuroki from the Japanese Society of Phlebology for technical support. The authors attest they are in compliance with human studies committees and animal welfare regulations of the authors’ institutions and Food and Drug Administration guidelines, including patient consent where appropriate. For more information, visit the Author Center. Appendix For supplemental tables, please see the online version of this paper. ==== Refs References 1 Huang C. Wang Y. Li X. Clinical features of patients infected with 2019 novel coronavirus in Wuhan, China Lancet 395 2020 497 506 31986264 2 Guan W.J. Ni Z.Y. Hu Y. Clinical characteristics of coronavirus disease 2019 in China N Engl J Med 382 2020 1708 1720 32109013 3 Coronavirus resource center Johns Hopkins University of Medicine https://coronavirus.jhu.edu/ 4 Ma L. Song K. Huang Y. Coronavirus disease-2019 (COVID-19) and cardiovascular complications J Cardiothorac Vasc Anesth 35 2021 1860 1865 32451271 5 Lodigiani C. Iapichino G. Carenzo L. Venous and arterial thromboembolic complications in COVID-19 patients admitted to an academic hospital in Milan, Italy Thromb Res 191 2020 9 14 32353746 6 Poissy J. Goutay J. Caplan M. Pulmonary embolism in patients with COVID-19: awareness of an increased prevalence Circulation 142 2020 184 186 32330083 7 Zhang L. Feng X. Zhang D. Deep vein thrombosis in hospitalized patients with Covid-19 in Wuhan, China: prevalence, risk factors, and outcome Circulation 142 2020 114 128 32421381 8 Piazza G. Campia U. Hurwitz S. Registry of arterial and venous thromboembolic complications in patients with COVID-19 J Am Coll Cardiol 76 2020 2060 2072 33121712 9 Ackermann M. Verleden S.E. Kuehnel M. Pulmonary vascular endothelialitis, thrombosis, and angiogenesis in Covid-19 N Engl J Med 383 2020 120 128 32437596 10 Thachil J. Tang N. Gando S. ISTH interim guidance on recognition and management of coagulopathy in COVID-19 J Thromb Haemost 18 2020 1023 1026 32338827 11 Cuker A. Tseng E.K. Nieuwlaat R. American Society of Hematology 2021 guidelines on the use of anticoagulation for thromboprophylaxis in patients with COVID-19 Blood Adv 5 2021 872 888 33560401 12 Connors J.M. Brooks M.M. Sciurba F.C. Effect of antithrombotic therapy on clinical outcomes in outpatients with clinically stable symptomatic COVID-19: the ACTIV-4B randomized clinical trial JAMA 326 2021 1703 1712 34633405 13 Yamashita Y. Maruyama Y. Satokawa H. Incidence and clinical features of venous thromboembolism in hospitalized patients with coronavirus disease 2019 (COVID-19) in Japan Circ J 85 2021 2208 2214 34011824 14 Yamashita Y. Yamada N. Mo M. The primary prevention of venous thromboembolism in patients with Covid-19 in Japan: current status and future perspective Ann Vasc Dis 14 2021 1 4 33786092 15 Nishimoto Y. Yachi S. Takeyama M. The current status of thrombosis and anticoagulation therapy in patients with COVID-19 in Japan: from the CLOT-COVID study J Cardiol 80 4 2022 285 291 35430141 16 Sakamoto J. Yamashita Y. Morimoto T. Cancer-associated venous thromboembolism in the real world- from the COMMAND VTE registry Circ J 83 2019 2271 2281 31548438 17 Schulman S. Kearon C. Subcommittee on Control of Anticoagulation of the Scientific and Standardization Committee of the International Society on Thrombosis and Haemostasis. Definition of major bleeding in clinical investigations of antihemostatic medicinal products in non-surgical patients J Thromb Haemost 3 2005 692 694 15842354 18 Thygesen K. Alpert J.S. Jaffe A.S. Third universal definition of myocardial infarction Circulation 126 2012 2020 2035 22923432 19 Iba T. Levy J.H. Levi M. Connors J.M. Thachil J. Coagulopathy of coronavirus disease 2019 Crit Care Med 48 2020 1358 1364 32467443 20 Cui S. Chen S. Li X. Liu S. Wang F. Prevalence of venous thromboembolism in patients with severe novel coronavirus pneumonia J Thromb Haemost 18 2020 1421 1424 32271988 21 Klok F.A. Kruip M. van der Meer N.J.M. Confirmation of the high cumulative incidence of thrombotic complications in critically ill ICU patients with COVID-19: An updated analysis Thromb Res 191 2020 148 150 32381264 22 Leentjens J. van Haaps T.F. Wessels P.F. Schutgens R.E.G. Middeldorp S. COVID-19-associated coagulopathy and antithrombotic agents-lessons after 1 year Lancet Haematol 8 2021 e524 e533 33930350 23 Kim H.K. Tantry U.S. Park H.W. Ethnic difference of thrombogenicity in patients with cardiovascular disease: a pandora box to explain prognostic differences Korean Circ J 51 2021 202 221 33655720 24 Kim H.K. Tantry U.S. Smith S.C. Jr. The East Asian paradox: an updated position statement on the challenges to the current antithrombotic strategy in patients with cardiovascular disease Thromb Haemost 121 2021 422 432 33171520 25 Bhatla A. Mayer M.M. Adusumalli S. COVID-19 and cardiac arrhythmias Heart Rhythm 17 2020 1439 1444 32585191 26 Sala S. Peretto G. De Luca G. Low prevalence of arrhythmias in clinically stable COVID-19 patients Pacing Clin Electrophysiol 43 2020 891 893 32543745 27 Libby P. Luscher T. COVID-19 is, in the end, an endothelial disease Eur Heart J 41 2020 3038 3044 32882706 28 Romiti G.F. Corica B. Lip G.Y.H. Proietti M. Prevalence and impact of atrial fibrillation in hospitalized patients with COVID-19: a systematic review and meta-analysis J Clin Med 10 2021 2490 34199857 29 Adomi M. Kuno T. Komiyama J. Association between pre-admission anticoagulation and in-hospital death, venous thromboembolism, and major bleeding among hospitalized COVID-19 patients in Japan Pharmacoepidemiol Drug Saf 31 2022 680 688 35324035 30 Lopes R.D. de Barros E.S.P.G.M. Furtado R.H.M. Therapeutic versus prophylactic anticoagulation for patients admitted to hospital with COVID-19 and elevated D-dimer concentration (ACTION): an open-label, multicentre, randomised, controlled trial Lancet 397 2021 2253 2263 34097856 31 Fujiwara S. Nakajima M. Kaszynski R.H. Prevalence of thromboembolic events and status of prophylactic anticoagulant therapy in hospitalized patients with COVID-19 in Japan J Infect Chemother 27 2021 869 875 33663933 32 Oba S. Hosoya T. Amamiya M. Arterial and venous thrombosis complicated in COVID-19: a retrospective single center analysis in Japan Front Cardiovasc Med 8 2021 767074 33 Horiuchi H. Morishita E. Urano T. Yokoyama K. Questionnaire-survey Joint Team on the COVID-19-related Thrombosis. COVID-19-related thrombosis in Japan: final report of a questionnaire-based survey in 2020 J Atheroscler Thromb 28 2021 406 416 33678766 34 Jimenez D. Garcia-Sanchez A. Rali P. Incidence of VTE and bleeding among hospitalized patients with coronavirus disease 2019: a systematic review and meta-analysis Chest 159 2021 1182 1196 33217420 35 Helms J. Tacquard C. Severac F. High risk of thrombosis in patients with severe SARS-CoV-2 infection: a multicenter prospective cohort study Intensive Care Med 46 2020 1089 1098 32367170 36 Fraisse M. Logre E. Pajot O. Mentec H. Plantefeve G. Contou D. Thrombotic and hemorrhagic events in critically ill COVID-19 patients: a French monocenter retrospective study Crit Care 24 2020 275 32487122 37 Spyropoulos A.C. Goldin M. Giannis D. Efficacy and safety of therapeutic-dose heparin vs standard prophylactic or intermediate-dose heparins for thromboprophylaxis in high-risk hospitalized patients with COVID-19: the HEP-COVID randomized clinical trial JAMA Intern Med 181 2021 1612 1620 34617959 38 Sholzberg M. Tang G.H. Rahhal H. Effectiveness of therapeutic heparin versus prophylactic heparin on death, mechanical ventilation, or intensive care unit admission in moderately ill patients with covid-19 admitted to hospital: RAPID randomised clinical trial BMJ 375 2021 n2400 34649864

0

PMC9753959

NO-CC CODE

2022-12-16 23:26:41

no

JACC Asia. 2022 Dec 15; 2(7):897-907

utf-8

JACC Asia

10.1016/j.jacasi.2022.09.005

oa_other

==== Front Lancet Lancet Lancet (London, England) 0140-6736 1474-547X Elsevier Ltd. S0140-6736(22)00162-3 10.1016/S0140-6736(22)00162-3 Comment The precariousness of balancing life and death Grant Liz a Khan Farzana b a Global Health Academy, University of Edinburgh, Edinburgh EH8 9AG, UK b Fasiuddin Khan Research Foundation, Uttara, Dhaka, Bangladesh 1 2 2022 26 February-4 March 2022 1 2 2022 399 10327 775777 © 2022 Elsevier Ltd. All rights reserved. 2022 Elsevier Ltd Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. ==== Body pmcDespite the centrality of death to our lives, people from many societies avoid meaningful conversations about death, and its value as a fundamental human experience has been largely lost. Diminishing the inevitability and humanity of death has obscured our understanding of health and life. Dying in the 21st century is, as highlighted in the Lancet Commission on the Value of Death,1 a story of paradox. Advances in technologies, science, medicine, artificial intelligence, and pharmaceuticals have saved lives but have complicated death in high-resourced health systems. Many people today die after substantial efforts at what is often called futile care. Such overtreatment in hospitals, mostly serving those with higher socioeconomic status, contrasts with a great global abyss of undertreatment.2 From the perspectives of those living in countries without adequate health resources, dying is too often characterised by gross inequity in access to basic care or support. More than 61 million people globally experience serious and avoidable health-related pain and suffering, and many people continue to die from preventable illnesses.3 The poorest 50% of the world's population live in countries that have only 1% of the distributed morphine equivalent medication essential to alleviating pain.4 Is the Lancet Commission on the Value of Death relevant in countries where the challenge is to constantly balance dying and death in poverty and inequity? In countries where universal health coverage is missing, multiple factors can determine death and dying. These include insufficient services, resources, training, or drugs; the cost of accessing care; reluctance of health staff to break bad news because of cultural, social, and time pressure reasons; not receiving the right level of care with costly interventions; barriers in getting to the right or the safe place to die because of siloed health services or the absence of basics such as sheets, mattresses, hygiene, and running water. All these factors can be addressed. Unlike so many other priorities in global health, affordability is not the greatest barrier for all countries to deliver services of care to die well; valuing those who are dying is. Dying has become one of the costliest health-care events. Spending in the last year of life accounts for a disproportionate share of total health expenditure in high-income countries notes the Commission. The global gap in services and therapies does not mean death or dying is cheap for people living in low-income countries. Inadequate investment in effective palliative care interventions in these settings contributes to intergenerational poverty, with children taken out of school when savings for education are used instead for care.5 The Commission draws parallels between the need to rebalance our relationship with death with that of balancing our relationship to the planet. The climate crisis, ecosystem collapse, and biodiversity loss are not only causing untimely deaths, but also point towards planetary death. Ill health and death have been brought closer by the direct and indirect impacts of climate on health, but as more demands are put on the health sector the more the sector becomes a driver of the climate crisis.6 Is this fractured relationship with nature connected to a detachment from death and the way societies have deluded themselves into believing both nature and death can be manipulated, tamed, and managed? The Commission uses the construct of “death systems” to explore the complex components that determine how care of the dying and the dead is given, who is included and excluded from such care, where care happens, and the dynamic shift in who “owns” death. 30 recommendations are made to bring about radical change in death systems, acknowledging that death systems are unique to societies, shaped by culture, history, religious beliefs, and resources. But what happens when the death of those who speak out against the state or who are members of minority communities is pursued by a political regime, or when dying is on such a scale that the death system breaks down, such as in Yemen, and Syria, and increasingly in Afghanistan?7, 8, 9 With the 21st century expected to be a century of mass migration, lessons need to be learned from refugees, internally displaced peoples, women and girls who are trafficked, and persecuted communities who are balancing life and death in fragility. The experiences and priorities of these groups do not feature in the Commission. Does death hold a different value in contexts of political oppression or migration? Or of entrenched racism, xenophobia, or misogyny including femicide? The anniversary of the 2021 military coup in Myanmar on Feb 1, 2022 is a stark reminder that health workers have been the target of violence and death by the military regime.10, 11 The Rohingya community in Cox's Bazar, Bangladesh, and refugees making crossings of the Mediterranean have had to develop ways of facing death and caring for the dead and dying which sit outside the surrounding societal structures and norms.12, 13 In these cases death and life will struggle to be “rebalanced” as the Commission proposes. © 2022 Pakri/500px/Getty Images 2022 As the Commission highlights, death occurs through conflict, accident, natural disaster, pandemic, violence, suicide, neglect, or disease. The World Economic Forum Global Risks Report 2022 identifies the ten most severe risks over the next 10 years, including livelihood and debt crisis, severe weather, infectious disease, environmental damage, and geo-economics confrontation, and points to a world where there will rarely be a singular cause of death.14 Perhaps the greatest challenge societies face in repositioning death systems will be to move from siloed sectors into interconnected ones. In so many societies we have lost trust in, and relegated, our ability to deal with death. The medicalisation of death and the capability or otherwise of a health system to manage death has come to determine the way that death is treated. The Commission argues that only by re-establishing the value of death will we be able to transform our health systems. The Commission offers a vision of a new system for death and dying underpinned by five principles—tackling the social determinants of death, dying, and grieving; seeing death as a relational and spiritual process; enabling networks of informal and formal care; normalising conversations and stories of death, dying, and grief; and recognising death has a value. This framing points to ways to improve the experience of death and dying globally. Achievement of the Commission's vision will require a renewed belief in a shared humanity and the recognition that we are born equal, but into very unequal circumstances, and although we cannot change the inevitability of death, societies can change the circumstances to avert preventable deaths and provide the time, space, comfort, and compassion to die. We declare no competing interests. ==== Refs References 1 Sallnow L Smith R Ahmedzai SH Report of the Lancet Commission on the Value of Death: bringing death back into life Lancet 2022 published online Jan 31. 10.1016/S0140-6736(21)02314-X 2 Khan F Ahmad N Iqbal M Kamal AM Physicians' knowledge and attitude of opioid availability, accessibility and use in pain management in Bangladesh Bangladesh Med Res Council Bull 40 2014 18 24 3 Knaul FM Farmer PE Krakauer EL Alleviating the access abyss in palliative care and pain relief—an imperative of universal health coverage: the Lancet Commission report Lancet 391 2017 1391 1454 29032993 4 Bhadelia A De Lima L Arreola-Ornelas H Kwete XJ Rodriguez NM Knaul FM Solving the global crisis in access to pain relief: lessons from country actions Am J Public Health 109 2019 58 60 30495996 5 Anderson RE Grant L What is the value of palliative care provision in low-resource settings? BMJ Glob Health 2 2017 e000139 6 Watts N Amann M Arnell N The 2019 report of The Lancet Countdown on health and climate change: ensuring that the health of a child born today is not defined by a changing climate Lancet 394 2019 1836 1878 31733928 7 Looi MK Covid-19: deaths in Yemen are five times global average as healthcare collapses BMJ 370 2020 m2997 8 Blanchet K Fouad FM Pherali T Syrian refugees in Lebanon: the search for universal health coverage Confl Health 10 2016 12 27252775 9 UNHCR UN and partners launch plans to help 28 million people in acute need in Afghanistan and the region https://www.unhcr.org/news/press/2022/1/61dc5d024/un-partners-launch-plans-help-28-million-people-acute-need-afghanistan.htm Jan 11, 2022 10 Soe Z Oo M Wah K Naing A Skalicky-Klein R Phillips G Myanmar's health leaders stand against military rule Lancet 397 2021 875 11 Dyer O Myanmar coup regime targets striking doctors with bullets, arrests, and erasures BMJ 373 2021 n1076 12 Doherty M Power L Petrova M Illness-related suffering and need for palliative care in Rohingya refugees and caregivers in Bangladesh: a cross-sectional study PLoS Med 17 2020 e1003011 13 Last T Spijkerboer T Tracking deaths in the Mediterranean Brian T Laczko F Fatal journeys: tracking lives lost during migration 2014 International Organization for Migration Geneva 85 106 14 World Economic Forum World Economic Forum Global Risks Report https://www.weforum.org/reports/global-risks-report-2022 2022

35114145

PMC9753972

NO-CC CODE

2022-12-16 23:26:23

no

Lancet. 2022 Feb 1 26 February-4 March; 399(10327):775-777

utf-8

Lancet

10.1016/S0140-6736(22)00162-3

oa_other

==== Front Lancet Lancet Lancet (London, England) 0140-6736 1474-547X Elsevier Ltd. S0140-6736(22)00146-5 10.1016/S0140-6736(22)00146-5 Perspectives Ros Taylor: seeing palliative care as relational Samarasekera Udani 1 2 2022 26 February-4 March 2022 1 2 2022 399 10327 783783 © 2022 Elsevier Ltd. All rights reserved. 2022 Elsevier Ltd Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. ==== Body pmc While studying sociology as a medical student at Cambridge University, UK, Ros Taylor wrote her dissertation on the social status of the dying. In 1977, 10 years after the first modern hospice opened in the UK, “I was writing about institutionalisation of the dying…and the need for more open awareness of dying”, she recalls. This interest in care at the end of life stayed with her. Taylor is now Strategic Medical Lead at Harlington Hospice and Michael Sobell Hospice, London, UK, and Palliative Lead at Hillingdon Health and Care Partners, a partnership to provide joined-up care for people in west London. She is also a Commissioner for the Lancet Commission on the Value of Death, which sets out a new vision of death and dying and principles to achieve it. “The Commission is looking at changing the industrialisation of medicine that's happened”, says Taylor. “Medicine can do so much now. What we see in hospital is more and more interventions are done, but it's no longer as thoughtful as it could be, and the essential conversations of what matters to people are not taking place. It is the understanding that we are mortal and have a finite time on this planet that needs to be revisited in medicine.” Taylor's engagement with how we die has defined her career. But she also has personal experience of the impact of a family death. Taylor faced devastating loss when her daughter Phoebe, aged 31 years, died by suicide in 2021. Taylor decided to share her grief on social media and posts thoughts and memories about Phoebe. “Not just to remember her, but also to encourage other people to share their stories. It's certainly helped me to capture the essence of Phoebe, and to challenge the stigma of suicide”, she says. The way Taylor shared this loss points to her openness in talking about death more generally, a feature of her work in palliative care. After her medical degree at Cambridge and Westminster Medical School, Taylor's interest in palliative care grew when she worked as a general practitioner in Cambridgeshire and Cumbria. “I loved the holistic style of general practice” that existed then, she says. During her decade in general practice, she became drawn towards caring for patients with terminal illnesses and their families. But it was after hearing a lecture by Fiona Randall, a UK pioneer of palliative medicine, that Taylor realised she wanted to work in the specialty. After setting up a small palliative care unit in West Cumberland Hospital, Cumbria, Taylor decided to leave general practice and pursue a career in palliative and hospice medicine. She was appointed medical director of the Hospice of St Francis, Berkhamsted, in 1996 and realised early on that the hospice needed a different building. “I was having serious conversations with families in the bathroom because there was nowhere else to sit down and talk”, she recalls. She devoted the next few years to developing a multidisciplinary medical team and building a new hospice, which opened in 2007. In 2015, she became clinical director at the national charity Hospice UK, a role which included encouraging best practices to be shared and advocating for better funding for hospices. It “opened up a whole new perspective exploring the charitable role; why end-of-life care is funded predominately by charity in the UK and the problems that go with that. I used to rant in lectures about why care of people who are dying was funded by selling second-hand clothes and running marathons, which it still is”, she says. While in this role and missing clinical work, Taylor became an honorary consultant in palliative care at the Royal Marsden and Royal Brompton Hospitals, London. In these leadership roles Taylor has been a change-maker and supported her colleagues, according to Max Watson, palliative medicine consultant at Western Trust, Enniskillen, UK. “Ros has made good things happen wherever she has worked and has been willing to put in the effort and deal with the organisational challenges to make it happen…As medical director of a hospice for many years she provided leadership by inspiration. She leads by encouraging people to be more than they could ever expect themselves to be and by affirming people's potential,” Watson comments. Taylor has a special interest in a holistic approach to palliative medicine, first inspired by the 2000 International Congress on Palliative Care, Montreal, QC, Canada. This holistic approach, which she teaches to trainee doctors, “means understanding wherever somebody is in their life, even if they are only days away from death, they are a person with things that matter to them, that they have a past and a present and possibly a future, however short that might be, and we need to understand their priorities. So holistic care has always been about seeing palliative care as relational rather than transactional and seeing the patient as a part of a network within their family and their community”, she says. To improve end-of-life care, Taylor believes palliative care should be a larger part of training for all health professionals. “Many medical school curricula still have just one day in 5 years focusing on palliative care issues, often with no placements in a hospice or palliative care team”, she notes. Taylor thinks it is a priority for palliative care to be reinstated into primary care in the UK. “I believe that palliative medicine is just good family medicine with more time”, she says. “Most of what we do is talking to people, discovering what and who matters, and seeing if we can facilitate those priorities, those goals of care”, she explains. In her hospice work, Taylor and her team recently helped a 93-year-old man reach his goal of care: playing the piano again. She reflects: “I now see our hospice as a fixer, trying to create memories for the people who are living on. We do an awful lot of that.”

35114147

PMC9753973

NO-CC CODE

2022-12-16 23:26:23

no

Lancet. 2022 Feb 1 26 February-4 March; 399(10327):783

utf-8

Lancet

10.1016/S0140-6736(22)00146-5

oa_other

==== Front Lancet Lancet Lancet (London, England) 0140-6736 1474-547X Elsevier Ltd. S0140-6736(22)00891-1 10.1016/S0140-6736(22)00891-1 Comment Effective post-pandemic governance must focus on shared challenges Williamson Anne a Forman Rebecca b Azzopardi-Muscat Natasha c Battista Robert d Colombo Francesca e Glassman Amanda f Marimont Josep Figueras g Javorcik Beata h O'Neill Jim i McGuire Alistair b McKee Martin j Monti Mario k O'Donnell Gus l Wenham Clare b Yates Robert i Davies Sally m Mossialos Elias b a Barts and the London School of Medicine and Dentistry, London E1 2AD, UK b London School of Economics and Political Science, London, UK c WHO Regional Office for Europe, Copenhagen, Denmark d Dr Evidence, Santa Monica, CA, USA e Organisation for Economic Co-operation and Development, Paris, France f Center for Global Development, Washington, DC, USA g European Observatory on Health Systems and Policies, WHO European Centre for Health Policy, Eurostation (Office 07C020), Brussels, Belgium h European Bank for Reconstruction and Development, London, UK i Chatham House, London, UK j Department of Health Services Research and Policy, London School of Hygiene & Tropical Medicine, London, UK k Bocconi University, Milan, Italy l Frontier Economics, London, UK m Trinity College, University of Cambridge, Cambridge, UK 16 5 2022 28 May-3 June 2022 16 5 2022 399 10340 19992001 © 2022 Elsevier Ltd. All rights reserved. 2022 Elsevier Ltd Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. ==== Body pmcThe COVID-19 pandemic has highlighted profound weaknesses in the global governance of health; inadequate preparation, coordination, and accountability hampered the collective response of nations at each stage. Changes to the global health architecture are necessary to mitigate the health and socioeconomic damage of the ongoing pandemic, and to prepare for the next major global threat to health. Against this backdrop, on April 4, 2022, the London School of Economics and Political Science, London, UK, hosted a meeting on the topic, “Paying the Pandemic Piper: Global Health and Economic Security”. The cross-sectoral stakeholders who participated at the meeting arrived at several insights, including the key proposals captured here. We recommend international institutions focus on their core missions and unique capabilities to respond to global externalities—ie, policy areas and challenges where the actions or inaction of any one country affect all global actors. Within the multilateral space there are many overlapping, fragmented efforts to improve global governance in response to COVID-19, with different actors leading each process. Examples of proposals to improve pandemic prevention, preparedness, and response are shown in the panel .1, 2, 3, 4, 5 Panel Key proposals to improve global governance of health and pandemic prevention, preparedness, and response World Health Assembly: development of a pandemic treaty • A special session of the World Health Assembly on Dec 1, 2021 established an intergovernmental body to negotiate a forward-focused international instrument, a so-called pandemic treaty, to strengthen pandemic prevention, preparedness, and response1 WHO: consultation on emergency pandemic preparedness • WHO is consulting member states on a range of preparedness and response proposals, including a Global Health Emergency Council and scaled-up Universal Health and Preparedness reviews2 G20 and World Bank: Financial Intermediary Fund and G20 Finance and Health Board • G20 members, led by Indonesia, Italy, and the USA, are championing a Financial Intermediary Fund for pandemic preparedness and response to be hosted by the World Bank Group3 • A G20 Finance and Health Board has also been proposed to enhance coordination between finance and health strategies towards a stronger health security architecture, building on the G20 Joint Finance and Health Task Force4 USA and European Commission: changes to the International Health Regulations • The USA and the European Commission are proposing reform to the International Health Regulations to make them more relevant and move beyond the current name and shame of individual countries5 The 75th World Health Assembly in Geneva, Switzerland, on May 22–28, 2022 is an opportunity to turn these varied proposals into effective action. Success will require clear demarcations of responsibility with a parsimonious role for international institutions. Although a pandemic treaty or alternative new instrument or process cannot solve all that is wrong with global health, it can deliver targeted improvements if supported by effective and clear global governance. Such a new governance instrument should focus on responsibilities that are truly transnational in nature, meaning it addresses global challenges with causes and consequences that transcend national boundaries. Importantly, the processes summarised in the panel need to unite these international actors and strengthen their collective efforts rather than reinforce historical silos in global health governance. This collective response means that WHO pandemic treaty negotiations should align with the G20 and World Bank proposals of a Financial Intermediary Fund and a Finance and Health Board.6 Additionally, many countries are striving to increase domestic health spending, despite the economic damage from the impacts of the COVID-19 pandemic.7 More should be done to align this momentum for increased health spending to make best use of national government resources. Our core recommendation is for international institutions to focus on fulfilling their unique capabilities by sharing knowledge between countries; pooling resources and distributing benefits equitably between countries; monitoring the preparedness of health systems within and across countries; and convening national actors effectively and in real time. First, international institutions should synthesise, create, and share knowledge of which health policies work and which do not, rather than trying to duplicate or mandate national actions. For example, different national lockdown policies have had widely divergent benefits and costs, yet there is no repository of global comparative data to analyse the impacts of lockdowns to date.8 The lockdowns and economic impacts of the COVID-19 pandemic have underlined the interdependence between health and economic security. National decision makers need a common language of outcomes to compare the trade-offs within and across sectors. International institutions could support this reframing of interdependent outcomes, with health spending explicitly segmented into health maintenance and health investment so that new opportunities are easily identified.9 Second, international institutions can pool resources, since aspects of pandemic preparedness and response benefit from economies of scale. For example, platform technologies that enable vaccine research and development could be pooled under a clear global investment framework with guaranteed supply to contributors alongside global equity provisions—although the sustainability of these investments also requires attention on the demand side.10 Economies of scale are also relevant for public health assets, such as high complexity laboratories, regulatory harmonisation, and medical supplies procurement, with potential for mutually beneficial collaboration. Third, shared monitoring of the preparedness of national health systems for crises is required, following the model of the G20 Financial Stability Board, which tests financial system resilience so that weaknesses can be addressed.11 Such a role is suited to a single international body to ensure consistency and transparency, and could be attained by expanding the WHO pilot Universal Health and Preparedness reviews.2 The International Monetary Fund could also regularly analyse the economic and financial consequences of health challenges, as an input to the proposed G20 Finance and Health Board. Finally, international institutions have the unique power to convene health, finance, and political actors, but this has not always been done well historically. The proposed G20 Finance and Health Board would strengthen coordination between finance and health, which would be a promising step.4 However, more could be done in regular forums to strengthen collaboration. One relevant model here is the UN Conference of the Parties, whereby countries make climate pledges and reassess them annually. Participants could then derive financial and non-financial support from parties with mutual benefit. Convening key global actors would also encourage these forums to act on unmet health and economic challenges. Given the potential for disease outbreaks and antimicrobial resistance (AMR) to spread across borders, collaborative action offers regional and global benefits. Improvements to global governance structures for pandemic preparedness and response will also have benefits for other global health challenges, such as climate change and mass migration.12 Prevention and detection activities are integral to tackling future pandemics and AMR, and global governance should maximise the shared benefits across these two areas. Health leaders must not squander this opportunity to build a renewed global and national architecture to meet current challenges and those yet to come. The London School of Economics and Political Science, London, UK, hosted a meeting of cross-sectoral stakeholders on the topic, “Paying the Pandemic Piper: Global Health and Economic Security”. We declare no competing interests. The views expressed in this Comment are entirely those of the authors and do not necessarily represent the views of the organisations with which they are affiliated. Participants at the meeting also included Chikwe Ihekweazu, Paul Johnson, Ilona Kickbusch, Connor Rochford, and Minouche Shafik. We thank George Wharton for his assistance with preparing this Comment. ==== Refs References 1 WHO World Health Assembly agrees to launch process to develop historic global accord on pandemic prevention, preparedness and response https://www.who.int/news/item/01-12-2021-world-health-assembly-agrees-to-launch-process-to-develop-historic-global-accord-on-pandemic-prevention-preparedness-and-response Dec 1, 2021 2 WHO Strengthening the global architecture for health emergency preparedness, response and resilience. White paper for consultation https://www.who.int/publications/m/item/white-paper-consultation-strengthening-the-global-architecture-for-health-emergency-preparedness-response-and-resilience May 4, 2022 3 Yellen J Ghebreyesus T Indrawati S The next pandemic doesn't have to hit so hard. Foreign Policy https://foreignpolicy.com/2022/04/21/pandemic-covid-19-future-global-health-security-architecture/ April 21, 2022 4 Monti M Torbica A Mossialos E McKee M A new strategy for health and sustainable development in the light of the COVID-19 pandemic Lancet 398 2021 1029 1031 34516953 5 Voss M Wenham C Eccleston-Turner M Sangameshwaran R Detering B A new pandemic treaty: what the World Health Organization needs to do next March 30, 2022 London School of Economics and Political Science https://blogs.lse.ac.uk/covid19/2022/03/30/a-new-pandemic-treaty-what-the-world-health-organization-needs-to-do-next/ 6 World BankWHO Analysis of pandemic preparedness and response (PPR) architecture and financing needs and gaps https://g20.org/wp-content/uploads/2022/02/G20-FHTF-Financing-Gaps-for-PPR-WHOWB-Feb-10_Final.pdf 2022 7 Economist Intelligence Unit Covid-19: the impact on healthcare expenditure 2020 The Economist https://www.eiu.com/n/campaigns/covid-19-the-impact-on-healthcare-expenditure/#mktoForm_anchor 8 Oraby T Tyshenko MG Maldonado JC Modeling the effect of lockdown timing as a COVID-19 control measure in countries with differing social contacts Sci Rep 11 2021 3354 9 Pan-European Commission on Health and Sustainable Development Drawing light from the pandemic: a new strategy for health and sustainable development https://www.euro.who.int/en/health-topics/health-policy/european-programme-of-work/pan-european-commission-on-health-and-sustainable-development/publications/drawing-light-from-the-pandemic-a-new-strategy-for-health-and-sustainable-development-2021 2021 10 Forman R Shah S Jeurissen P Jit M Mossialos E COVID-19 vaccine challenges: what have we learned so far and what remains to be done? Health Policy 125 2021 553 567 33820678 11 Financial Stability Board About the FSB https://www.fsb.org/about/ 2020 12 Anderson M Schulze K Cassini A Plachouras D Mossialos E A governance framework for development and assessment of national action plans on antimicrobial resistance Lancet Infect Dis 19 2019 e371 e384 31588040

35588759

PMC9754057

NO-CC CODE

2022-12-16 23:26:25

no

Lancet. 2022 May 16 28 May-3 June; 399(10340):1999-2001

utf-8

Lancet

10.1016/S0140-6736(22)00891-1

oa_other

==== Front Lancet Lancet Lancet (London, England) 0140-6736 1474-547X Elsevier Ltd. S0140-6736(22)00932-1 10.1016/S0140-6736(22)00932-1 Perspectives Impossible times Wilson Chloe 26 5 2022 28 May-3 June 2022 26 5 2022 399 10340 20082008 Farooki Roopa Everything is True: A Junior Doctor's Story of Life, Death and Grief in a Time of Pandemic2022Bloomsbury Publishing9781526633392 240£13·49© 2022 Elsevier Ltd. All rights reserved. 2022 Elsevier Ltd Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. ==== Body pmcDuring the COVID-19 pandemic health-care workers have often found themselves in the media spotlight. The pandemic has been framed as a war, and health-care workers have been labelled as soldiers on the front line. But using a wartime narrative to depict the pandemic is dangerous “None of you think you are soldiers, dodging spinning bullets shaped like pretty viral crowns”, asserts Roopa Farooki in her book Everything is True: A Junior Doctor's Story of Life, Death and Grief in a Time of Pandemic. Farooki writes with candid intimacy and eloquence about her experiences working as a newly qualified doctor in the UK's National Health Service (NHS) at the beginning of the pandemic. Farooki describes an unsettling atmosphere of chaos and frustration in the hospital. Patients were too scared to go to hospital because of what they had seen on the news. Those who did seek care in the hospital encountered inadequate resources, miscommunication, and unclear, ever-changing guidance. Beds in intensive care units and personal protective equipment were in short supply, doctors were being forced to make impossible decisions, rotas were changed to unsociable, long hours, and training was suspended. Despite the public clapping for “NHS heroes”, and politicians claiming to provide unprecedented support to the NHS, “You feel nothing. Just numb,” writes Farooki. The experience Farooki describes was similar to my own, working as a doctor at another UK hospital. Doctors, and all other health-care workers, have experienced great societal pressure and personal risk during the pandemic, with little attention to the protection of our rights. Health professionals have an absolute obligation to patients, but what about the duty of employers to protect health-care workers? And what happens when there is a conflict between our duty to care for patients and the duty to protect our own health and our family's health? In the UK, the General Medical Council defaults to doctors using their professional judgement in difficult circumstances and acknowledges that doctors might be required to work beyond the limits of their comfort zone. Farooki's Everything is True highlights the personal toll of the pandemic for doctors, and it is an important reminder that overstretching health-care workers and ignoring their wellbeing is ultimately detrimental to patients. Health professionals can choose to leave the NHS, patients are not afforded the same option. Everything is True is also a meditation on grief. Farooki reflects frequently on her close, yet competitive, relationship with her sister who died of breast cancer in the months before the pandemic. The loss seems to contribute to Farooki feeling overwhelmed and isolated: “You’re constantly on edge. You can’t physically sit down, unless you collapse to sleep. It weirds out your colleagues.” Home is not a retreat since there is tension with her husband, who is fearful of her job. “Not all lepers have spots. Some wear scrubs in A&E. You’re like Marie Curie playing with her radium”, writes Farooki. Patient safety depends on doctors’ wellbeing. Studies have shown that burnout increases the risks of making a major medical error. Yet there are still considerable barriers to seeking help within the profession, particularly in relation to mental illness, disabilities, and long-term conditions. Farooki comes into the hospital the day after her sister's funeral, but thankfully is sent home. The attitudes Farooki relates about physician sickness and wellbeing are saddeningly familiar to me. “You’re never ill, and you have always had an old-school sneaking suspicion that people who take sick days are frauds”, she writes. Yet it is not a weakness or a failure to become sick. Sickness presenteeism can compromise patient safety. One study suggests that trainees in particular were more likely to be motivated to work when sick to avoid appearing lazy or weak. Why are health professionals disinclined to follow the advice that we would presumably give to our patients? A deep sense of uncertainty pervades the ending of Everything is True. COVID-19 cases and deaths were still rising rapidly and no vaccine was yet available. Farooki describes how “Death and deterioration has been impossibly normalised. You’re living in impossible times.” Today, as the COVID-19 pandemic continues to evolve, impossible times seem far from over. Equitable access to COVID-19 diagnostics, treatments, and vaccines does not yet exist in many parts of the world. In the UK, despite a successful COVID-19 vaccination programme, the backlog of work in the NHS is immense. The number of people on surgical waiting lists in England is more than 6 million and the staffing crisis has deepened. To face this crisis and achieve good outcomes for patients, unacceptable workplace practices must be challenged and meaningful, institutional change is needed to support the wellbeing of both patients and the health workforce. ==== Refs Further reading Bianchi EF Bhattacharyya MR Meakin R Exploring senior doctors' beliefs and attitudes regarding mental illness within the medical profession: a qualitative study BMJ Open 6 2016 e012598 British Medical Association Disability in the medical profession. Survey findings 2020 2020 British Medical Association London Campbell D Staffing crisis deepens in NHS England with 110 000 posts unfilled The Guardian March 3, 2022 General Medical Council Coronavirus: your frequently asked questions https://www.gmc-uk.org/ethical-guidance/ethical-hub/covid-19-questions-and-answers#Working-safely 2022 Johnson SB Butcher F Doctors during the COVID-19 pandemic: what are their duties and what is owed to them? J Med Ethics 47 2021 12 15 33060186 Kaldijan LC Shinkunas LA Schacht Reisinger H Attitudes about sickness presenteeism in medical training: is there a hidden curriculum? Antimicrob Resist Infect Control 8 2019 149 31508227 The Lancet The NHS: the many challenges for leadership Lancet 398 2021 559 34391487

0

PMC9754058

NO-CC CODE

2022-12-16 23:26:25

no

Lancet. 2022 May 26 28 May-3 June; 399(10340):2008

utf-8

Lancet

10.1016/S0140-6736(22)00932-1

oa_other

==== Front Sci Total Environ Sci Total Environ The Science of the Total Environment 0048-9697 1879-1026 Elsevier B.V. S0048-9697(21)00181-9 10.1016/j.scitotenv.2021.145115 145115 Article Estimating the European CO2 emissions change due to COVID-19 restrictions Andreoni Valeria University of Liverpool, Management School, Chatham Street, Liverpool L69 7ZH, UK 14 1 2021 15 5 2021 14 1 2021 769 145115145115 16 11 2020 18 12 2020 7 1 2021 © 2021 Elsevier B.V. All rights reserved. 2021 Elsevier B.V. Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. The carbon dioxide variations generated by the socio-economic restrictions imposed by the management of the COVID-19 crisis are analysed in this paper for 23 European countries and 10 economic sectors. By considering the most up to date information on GDP and carbon intensity of production, this paper represents one of the first attempts to estimate the CO2 emissions change that have taken place in Europe during the first six months of 2020. Results show that more than 195,600 thousand tons of CO2 have been avoided between January and June 2020, compared to the same period of the previous year, representing a −12.1% emissions change. The largest reductions have taken place in the Manufacturing, Wholesale, Retail Trade, Transport, Accommodation and Food Service sectors, accounting for more than 93.7% of total CO2 change. Spain, Italy and France have been the most affected areas with −106,600 thousand tons emissions drop. In line with the results provided by previous studies, this paper highlights that the geographical and the sectoral distribution of the CO2 emissions change has been largely influenced by the magnitude of the COVID-19 impacts. In addition, the carbon intensity of production, characterizing the most affected economic activities, has been the main element of differentiation compared to the previous 2008 crisis. By providing preliminary estimation of the CO2 emissions change that have taken place across geographical and sectoral activities, this paper contributes to the existing climate policy debate and can support future estimation of CO2 variations both in a context of confinement release as well as in a context of reintroduced COVID-19 restrictions. Graphical abstract Unlabelled Image Keywords Lockdown Economic sectors Carbon intensity GDP variation Europe Editor: Jay Gan ==== Body pmc1 Introduction The outbreak of COVID-19 disease, firstly emerged in China at the end of December 2019, has caused unprecedented socio-economic disruptions and huge human life losses. According to data provided by the World Health Organization, over 44 million cases and more than 1.1 million deaths have been recorded during the first 10 months of the pandemic (WHO Covid-19 Dashboard). The rapid spread across developed and developing countries, together with the extensive pressure on the health systems have forced most of the governments to introduce various degree of social distance measures. By September 2020, at least 186 countries had imposed restrictions on movements with 82 of them affected by national or regional lockdown (Han et al., 2020). The forced shut-down of production and consumption activities together with reduced mobility and trade have however generated the deepest global recession since World War II (World Bank, 2020). The severity of the economic consequences and the related social instabilities have then induced most of the world countries to ease some of the restrictions, with the risk of additional pandemic waves. The large uncertainties related to the characteristics and treatment of the virus, together with the impossibility to forecast the duration of the crisis, makes it difficult to balance the trade-offs between health and socio-economic needs. Extensive debates have then been oriented to question the effectiveness of policies and to investigate multidimensional range of COVID-19 impacts (Arthi and Parman, 2020; Ibn-Mohammed et al., 2020). Despite the large attention devoted to economy and health (World Bank, 2020a) an increasing number of analysis has also been focusing on the environmental and climate effects (Sovacool et al., 2020). The possibility to use existing evidence and data to model and forecast the socio-economic and environmental impacts of reduced anthropogenic activities is representing a unique opportunity to discuss the existing sustainability constraints and to investigate opportunities for changes (Stoll and Mehling, 2020; Manzanedo and Manning, 2020; Shakil et al., 2020). Rume and Islam (2020), together with Zambrano-Monserrate et al. (2020) have for example provided preliminary overview of the COVID-19 effects highlighting the trade-offs existing between the environmental benefits of the reduced economic interactions and the environmental costs associated with consumption changes. The increased use of personal protective equipment and waste, together with the temporary reduction of recycling have for example been identified as some of the main environmental threats of the COVID-19 pandemic (Klemes et al., 2020). On the other side, the strict lockdown measures have contributed to drop the energy consumption rates with related impacts on emissions and air quality change. The forced reduction of transport activities and production has then represented an unprecedented opportunity to investigate the incidence of the anthropogenic emissions rate and an increasing number of studies have concentrated on that. Local variations of pollutants, such as PM2.5, PM10, CO, NO2, SO2, O3, and NH3 have for example been analysed for Indian cities (Sharma and Jain, 2020; Sharma et al., 2020; Mahato et al., 2020), Brazil (Nakada and Urban, 2020; Dantas et al., 2020), China (Li et al., 2020; Bao and Zhang, 2020; Isaifan, 2020; Wang et al., 2019), Europe (Menut et al., 2020; Tobia et al., 2020; Sicard et al., 2020; Ordonez et al., 2020; Baldasano, 2020; Collivignarelli et al., 2020) and United States (Berman and Ebisu, 2020; Chauhan and Singh, 2020). Most of these studies have used data from monitoring stations and satellite images, and have observed positive correlations between the severity of the lockdown restrictions and the air quality improvements. Analyses have also been concentrating on investigating the incidence of the lockdown measures into the greenhouse gases and the carbon dioxide generation. The lack of real time data and the global nature of these pollutants make however difficult to have direct information on the magnitude of change. For this reason, most of the existing studies have been using derivative approaches to estimate the countries and the sectoral emissions change. Recent attempts include Le Quere et al. (2020) that combining government policies and activity data forecasted an annual CO2 emissions drop ranging between −4% and −7%, depending on the duration of the restrictions. Liu et al. (2020) that used activity data from power generation, industry, transport and residential energy consumption to estimate a CO2 emissions change of −6.9% for China, −12% for Europe, and −9.5% for United States. The International Energy Agency (2020) that used fossil fuel energy demand to approximate a −5% CO2 emissions decline between January and April 2020. Han et al. (2021) that combined GDP changes and inventory data to calculate the sectoral and geographical CO2 emissions change for China, and Myllyvirta (2020) that used coal consumption and economic activity rates to estimate 18% CO2 emissions drop over a seven-week period in China. By considering the limited available information, these pioneering studies provide important contributions to understand the role that imposed socio-economic constraints can have in the carbon emissions trends and can support the definition of effective climate strategies. Most of the analysis conducted so far, have however been focused on the global or on the macro-regional sale (Le Quere et al., 2020) and no previous attempts have been specifically devoted to investigating the CO2 emissions change that have taken place in the different European countries. The European Environmental Agency is for example expecting to publish the first detailed report on greenhouse gases emissions in the autumn 2021 (EEA website). Given the primary role played by Europe in the international climate negotiations (EC website) and given the rapid changes induced by the ongoing COVID-19 crisis, the availability of timely estimation is of primary importance. Within this context, the main objective of this paper is to investigate the carbon dioxide emissions change that have taken place in Europe in the first six months of 2020, during which the most extensive lockdown restrictions have taken place. Using the most up to date information on GDP and carbon intensity of production, this study represents one of the first attempts to estimate the CO2 emissions change that have taken place in 23 European countries at industry level breakdown. By discussing the role that different economic activities are playing in the generation of carbon dioxide, the present paper can contribute to the existing climate policy debate. In addition, the provision of timely estimations can be functional to monitor and forecast the future CO2 emissions change both in a context of confinement release as well as in a context of reintroduced restrictions. 2 Data and methods Eurostat data have been used in this paper to estimate the carbon impacts of the COVID-19 outbreak for the 23 European countries for which consistent data are available. GDP data at industry level breakdown have been used to calculate the production variations that have taken place in the first semester of 2020 compared to the same period of the previous year. The carbon intensities provided by Eurostat for the year 2018, and reported in Table A1 of Appendix A, have been used to estimate the CO2 emissions change (Eurostat Air Emission database). In Table 1 , the industry breakdown considered in this paper is reported together with the related NACE Rev. 2 activity classification.1 Table 1 NACE Rev. 2 activity classification. Table 1Code Description A Agriculture, forestry and fishing C Manufacturing F Construction G, H, I Wholesale and retail trade, transport, accommodation and food service activities J Information and communication K Financial and insurance activities L Real estate activities M, N Professional, scientific and technical activities; administrative and support service activities O, P, Q Public administration, defence, education, human health and social work activities R, S, T, U Arts, entertainment and recreation; other service activities; activities of household and extra-territorial organizations and bodies Following the approach proposed by Han et al. (2021), where carbon factors and GDP variations are used to estimate the CO2 emission change, the carbon dioxide emissions drops are here estimated according to Eqs. (1), (2), where the carbon intensities of the NACE Rev.2 activities are assumed to remain unchanged from the year 2018.(1) CO2emissions=∑GDPit×EIi where GDP is the gross domestic product of time t, EI the carbon dioxide emissions intensity of production and i refers to the economic activities of Table 1. From Eq. (1), the carbon dioxide emissions changes are then estimated as:(2) ∆CO2emissions=∑Change rate ofGDPit×CO2emissions In the following section, results are discussed for the 23 European countries for which the most up to date and consistent data are available. 3 Results and discussion According to data reported in Table A2 of Appendix A, in the first semester of 2020, the CO2 emissions of the 23 European countries considered in this paper have been more than 195,600 thousand tons lower than in the same period of 2019, representing a percentage drop of 12.1%. Out of these, almost 106,600 thousand tons have taken place in Spain, Italy and France, that experienced the largest GDP (−12.7%, −12.1% and −11.9%, respectively) and CO2 emissions change (−22.5%, −18.2% and −16.5%, respectively). Heavily affected by the COVID-19 infections these countries have been among the first European areas to introduce lockdown restrictions (ECDC website). The closure of non-essential shops and production, the limitation imposed to the hospitality sector and the sharp decrease of movements, generated extensive implications across the entire economic compartments. The Wholesale, Retail Trade, Transport, Accommodation and Food Service Activities (G,H,I) together with the Manufacturing (C) sector, resulted to be the most affected activities, accounting for −61.7%% of the overall GDP changes. Traditionally characterized by some of the highest carbon intensities of production, these activities related to more than 93.7% of the total CO2 emissions drop. In Table A3 of Appendix A, the percentage contribution provided by every economic sector to the overall GDP and CO2 emissions change is reported for the 23 European countries considered in this paper. According to data provided by Eurostat, the motor vehicle, the textile and the furniture sectors resulted to be the most affected productions, while the pharmaceutical and tobacco have been the only compartments with positive growth rate (Eurostat industrial production database). Tourism related activities, such as food services and accommodation also performed some of the largest turnover decline, with an average 45.8% drop during the second quarter of 2020. Summer tourism destinations, such as Spain, Portugal and Italy recorded the largest percentage reduction (−78.2%, −65.9% and −62.6%, respectively), while Netherlands, Austria and Slovakia have been the only countries with a percentage drop lower than 15% (Eurostat services database). The Recreation (R, S, T, U), Professional Activities (M, N) and Public Administration (O, P, Q) have also been highly impacted by the COVID-19 restrictions, with an average GDP reduction of 30.6%. The low carbon intensities of these activities (Table A1 of Appendix A) have however generated minor CO2 emissions change (−5.4%). In most the European countries, the construction sector (F) has also been negatively affected, with exception of Germany, Sweden, Portugal, Netherland and some Eastern European areas, such as Poland, Romania and the Baltic states, that performed a GDP and carbon dioxide emission increase (Table A2 of Appendix A). Contrary to the decision taken by most of the other EU countries, these areas did not include the construction into the list of activities forced to shut down. Therefore, the impact on GDP has been lower than in countries, such as Italy, Luxemburg and France, that imposed more severe restriction to the construction compartment (IMF website; ECDC website). According to data provide by Eurostat, in these three countries the construction activities fell by 70.0%, 55.7% and 65.0% respectively, between February and April 2020, with reductions that have taken place both in the building and the civil engineering compartments (Eurostat construction database). When considering the carbon dioxide variation, the relatively low carbon intensity of construction (0.149 kg per euro, EU23 average) has induced a CO2 emission drop that accounted for less than 0.7% (−1343 thousand tons) of the total change. The Agricultural, Forestry and Fishing sector (A), has been the least affected economic compartment with a GDP variation accounting for 0.13% (−584 million euro) of the total change. According to data provide by OECD (2020a), the COVI-19 related spike in the consumption of vegetable and citrus fruits has contributed to boost a significant increase in the demand of agricultural products, with Spain experiencing the major sectoral gains (+587 million euro). In line with the analysis provided by Eurostat for the food-related retail sector, the demand of agricultural products increased during the two first months of the pandemic and stabilized in the following period (Eurostat retail trade database). The non-closure of supermarkets and essential shops, together with the panic buying behaviours has contributed to sustain the demand of the agricultural compartment (Jaspal et al., 2020). Consequently, minor variations have taken place for the related carbon dioxide emissions change, that reduced by 338 thousand tons (−0.17% of the total change). The Information and Communication (J) has been the second less affected sector, with a GDP reduction of 3713 million euro (−0.8% of the total GDP change). The working from home recommendations, the forced transition to telehealth and on-line educations, together with the rapid shifts to online retail have largely increased the demand for information systems technologies (He et al., 2020). As a consequence, 13 of the 23 European countries considered in this paper have accounted a GDP increase (Table A2 of Appendix A). Some of the largest variations have taken place in the Eastern European countries, such as Estonia, Lithuania and Romania, where the existing digitalization gaps have been partially reduced as a response to the induced digital transformation (World Bank, 2020b). When considering the carbon dioxide variation, the low carbon intensity of these productions has however generated minor emissions change, accounting for 0.01% of the total CO2 reduction. The Financial and Insurance (K) and the Real Estate Activities (L) have also been marginally affected by the COVID-19 related crisis, with an overall GDP drop of around 9800 million euro (2.1% of the total GDP change). As previously reported by OECD (2020b), the large set of measures introduced to support the investments of builders, lenders and tenants together with mortgage repayment suspension have contributed to reduce the overall losses and to promote an economic rebound in the period that followed the lockdown (EC, 2020). A detailed breakdown of the CO2 emissions change is reported in Fig. 1 according to sectoral and country level disaggregation.Fig. 1 CO2 emissions change (thousand tons) - industries and countries breakdown. Fig. 1 Despite the differences reported above, general trends can be identified across the industry and the geographical breakdown. As previously highlighted by other studied (Han et al., 2021; Sarkodie and Owusu, 2020; Rugani and Caro, 2020) the spatial distribution of the CO2 emissions change has been largely influenced by the magnitude of the COVID-19 impact. In addition, the emissions factors characterizing the economic activities of the different European countries have also influenced the variations. When the emission rate was higher 1 (kg per euro) the emission drop has been larger than the GDP change. The G,H,I sector of Denmark and Bulgaria, characterized by some of the highest emission factors (3.1 kg per euro and 2,3 kg per euro, respectively) have for example accounted for 98.4% and 71.2% of the respective CO2 drop, with a related 50.3% and 59.8% of GDP change (Table A3 of Appendix A). According to Eurostat data, in both countries, the carbon intensity of the transport sector (3.04 kg per euro in Denmark and 2.17 kg per euro in Bulgaria) is largely above the EU27 average (0.85 kg per euro) (Eurostat Air Emission database). In a similar way, the emissions rate of the manufacturing sector (C) of Bulgaria (1.12 kg per euro) and Cyprus (1.52 kg per euro) have generated an emission drop that has been higher than the GDP change. The opposite trend has been performed by Germany, Italy and Sweden, that have seen a GDP reduction largely higher than the carbon emission drop. According to data reported in Table A1 of Appendix A, these countries have the lowest carbon intensity of manufacturing (0.28 kg per euro, 0.33 kg per euro and 0.24 kg per euro, respectively). As previously highlighted by other studies (Han et al., 2021; Friedlingstein et al., 2019), the impacts on the most pollutant economic activities has been the main factor differentiating the present crisis from the previous 2008 crash, where the financial sector was the most affected economic compartment. The predominant role that the carbon intensities of production are playing in the estimations of the CO2 emissions change highlights the importance of up to date information. In relation to this paper, the existing data constraint related to the emissions factors of 2018, have probably generated a slightly overestimation of the CO2 emissions change. In addition, the unavailability of timely and consistent information related to the energy consumption changes that have taken places in offices and household spaces, have constrained the account of the carbon effects of the working from home recommendations. The limited breakdown of production activities and the related impossibility to provide disaggregated analysis for households and for some of the most affected economic sectors is also representing a limitation of the provided estimations. When consistent data will be available, detailed analysis should then be devoted to investigating the most affected economic activities, such as tourism and transport, together with the most impacted geographical areas. Up to now, the most consistent information related to transport has for example been published by Eurostat in relation to the number of passengers and tons of transported material (Eurostat website). The lack of specific data related to GDP changes, carbon intensities of passengers and freight transport activities, together with the limited information on distances and mobility indexes make however difficult to provide consistent estimations across the 23 European countries considered in this paper. From a methodological perspective, improvements could be achieved by using energy consumption as drivers of carbon emissions change, as recently attempted by Rugani and Caro (2020), Cheshmehzangi (2020) and Bulut (2020). The lack of timely and consistent energy data is however representing an operational constraint for the geographical and sectoral disaggregation considered in this paper. Despite the existing limitation, the analyses provided are a preliminary attempt to estimate the carbon dioxide emissions change in a context of limited information and dynamic changes. The results provided, together with the sectoral and the geographical breakdown, can then represent the base for additional analysis and research. Within this context, future investigations could for example be devoted to analysing the impacts of the expected rebound effect and to discuss the implication of potential behavioural changes. As previously highlighted by Sovacool et al. (2020) the transformed working habits, the increased investments in green mobility and the international cooperation emerged during the COVID-19 pandemic could represent a valuable opportunity for the management of the climate crisis. 4 Conclusion By considering a sectoral and geographical disaggregation of 10 economic compartments and 23 European countries, this paper estimates the CO2 emissions change that have taken place in the first six months of 2020 due to the imposed COVID-19 restrictions. Results show than more than 195,600 thousand tons of emissions have been avoided compared to the same period of the previous year. Spain, Italy and France, largely affected by the COVID-19 infection, have been the areas with the largest carbon reductions, accounting for more than 55% of the total emission changes. The Manufacturing, Wholesale Retail Trade, Transport Accommodation and Food Service sectors have been the most affected economic compartments and the relatively high carbon intensity of production has been the main element differentiating the carbon impacts of the present economic collapse from the previous 2008 crisis. The results provided are in line, even if slightly overestimated, with those of Liu et al. (2020) that calculated the sector-specific, country-level CO2 emissions change by using data from the international Carbon Monitor research initiative. When comparing the emission reductions for the industrial activities of Italy, France and Spain, the estimated results are higher in this paper than those provided by the carbon monitor database. The reason could be related to the types of activities included in the industrial sector, to the carbon intensities of fuel and to the fact that in the present study the emission factors of 2018 have assumed to be unchanged. Even with the existing limitations this paper represents one of the first attempts to estimate the national and the sectoral carbon dioxide emissions change that have taken place in Europe and can represent an initial platform for future analysis and researches. Despite the global impacts of the carbon dioxide emission changes, a better understanding of the regional and economic drivers of pollution is a fundamental element for the design of achievable climate solutions, as recently highlighted during the Paris climate negotiation. In this respect, the 8.5% of GDP drop and the related 12% carbon dioxide emissions change, that have taken place during the first 6 months of 2020, provide an indication of the carbon intensity of the existing economic structure. Within this context, the climate strategies defined in the recent COP agreements and the European objective of 40% greenhouse gas emission reduction should be questioned in relation to the possibility of replicating the emissions drop without causing the extreme socio-economic effects experienced in lockdown. As previously highlighted by other studies (Stoll and Mehling, 2020; Rume and Islam, 2020; Ferrarini et al., 2021) the carbon reduction auspicated by the international climate negotiation should be promoted in line with substantial transformation of the existing socio-economic structure. The extensive discussion emerging between the COVID-19 pandemic and the existing climate crisis should then investigate the carbon risks of the “return to normal” strategies and question the long-term sustainability solutions. The green recovery plans proposed by developed and developing countries, the behavioural changes induced on individual and societies, and the international cooperation emerged during the existing COVID-19 crisis is representing an important opportunity for change. After the failure of the sustainable recovery of the global financial crisis of 2008, the current pandemic can give us a chance that this time cannot be wasted. CRediT authorship contribution statement Valeria Andreoni: Conceptualization, Methodology, Investigation, Data analysis, Writing - original draft, Writing - review & editing, Visualization. Declaration of competing interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Appendix A Table A1 Carbon intensities – industry and country breakdown (kg per euro). Table A1 A C F G, H, I J K L M, N O, P, Q R, S, T, U Belgium 1.2100 0.6592 0.0951 0.6799 0.0209 0.0074 0.0026 0.0881 0.1157 0.3361 Bulgaria 0.5679 1.1245 0.2146 2.2957 0.0090 0.0013 0.0026 0.0750 0.0748 0.0283 Czechia 0.5210 0.3900 0.1406 1.0755 0.0089 0.0060 0.0048 0.1809 0.0555 0.0444 Denmark 0.8529 0.1665 0.1216 3.1110 0.0058 0.0049 0.0052 0.0479 0.0568 0.0447 Germany 0.5299 0.2801 0.1092 0.9167 0.0178 0.0147 0.0017 0.0386 0.0978 0.1053 Estonia 0.2754 0.5375 0.0781 1.3413 0.0059 0.0075 0.0399 0.0801 0.3025 0.1930 Spain 0.4179 0.5888 0.0067 1.1053 0.0101 0.0083 0.0005 0.0144 0.0713 0.0282 France 0.3726 0.3715 0.0846 0.5887 0.0085 0.0088 0.0018 0.0746 0.0909 0.1315 Italy 0.3025 0.3286 0.0784 0.6256 0.0045 0.0076 0.0023 0.0333 0.0723 0.0621 Cyprus 0.2129 1.5213 0.0905 0.4150 0.0076 0.0066 0.0017 0.0810 0.0430 0.0649 Latvia 0.7320 0.5151 0.1637 1.3474 0.0099 0.0123 0.0603 0.1092 0.1360 0.0859 Lithuania 0.4115 0.7310 0.0321 2.6628 0.0101 0.0120 0.0063 0.0457 0.1087 0.1029 Luxembourg 0.7634 0.5379 0.0560 2.0082 0.0086 0.0153 0.0082 0.0358 0.0854 0.1266 Hungary 0.5514 0.5375 0.2252 0.9284 0.0575 0.0503 0.0490 0.2035 0.2064 0.1518 Netherlands 0.7726 0.5677 0.0849 1.0101 0.0048 0.0088 0.0068 0.0666 0.0675 0.1505 Austria 0.3338 0.4026 0.1224 0.4205 0.0071 0.0035 0.0011 0.0266 0.0462 0.0640 Poland 2.2792 0.7971 0.0174 1.5737 0.0269 0.2403 0.0494 0.2020 0.3045 0.2525 Portugal 0.5716 0.6095 0.1991 0.9749 0.0089 0.0103 0.0009 0.0560 0.1024 0.1067 Romania 0.1741 0.5963 0.3701 0.8159 0.0264 0.0649 0.0315 0.1386 0.2120 0.2532 Slovenia 0.3436 0.3822 0.6003 0.6668 0.0524 0.0316 0.0196 0.2514 0.1291 0.2609 Slovakia 0.1201 0.8392 0.3543 0.9720 0.0116 0.0133 0.0346 0.1408 0.1726 0.3620 Finland 0.3467 0.3660 0.1058 1.1655 0.0008 0.0398 0.0063 0.0446 0.0680 0.0919 Sweden 0.3074 0.2446 0.0872 0.6924 0.0032 0.0045 0.0056 0.0438 0.0342 0.0499 Table A2 GDP and CO2 variations – industry and country breakdown (GDP: Million euro; CO2: Thousand tons). Table A2 A C F G, H, I J K L M, N O, P, Q R, S, T, U Total change % change Belgium GDP 52 −1813 −914 −6259 −271 −263 195 −2272 −2226 −738 −14,509 −7.53 CO2 63 −1195 −87 −4255 −6 −2 0 −200 −257 −248 −6187 −11.31 Bulgaria GDP −14 −399 −28 −483 85 47 −159 33 162 −51 −808 −4.19 CO2 −8 −449 −6 −1109 1 0 0 2 12 −1 −1558 −10.79 Czech Rep. GDP 69 −2530 −63 −1959 92 −100 −164 −196 17 −175 −5009 −6.17 CO2 36 −987 −9 −2107 1 −1 −1 −35 1 −8 −3110 −10.39 Denmark GDP 106 −509 −244 −2281 −7 259 −46 60 −895 −976 −4532 −3.59 CO2 91 −85 −30 −7096 0 1 0 3 −51 −44 −7210 −7.92 Germany GDP −55 −43,842 2417 −15,880 −1002 252 −15 −14,652 −12,198 −6040 −91,015 −6.52 CO2 −29 −12,279 264 −14,557 −18 4 0 −565 −1193 −636 −29,008 −8.18 Estonia GDP 8 −173 110 −263 113 28 −73 −27 21 −33 −289 −2.91 CO2 2 −93 9 −353 1 0 −3 −2 6 −6 −439 −8.81 Spain GDP 587 −11,333 −5536 −33,054 −1624 91 −1350 −7634 338 −5952 −65,467 −12.74 CO2 245 −6673 −37 −36,535 −16 1 −1 −110 24 −168 −43,269 −22.47 France GDP −469 −19,935 −11,226 −35,119 −2233 −3268 −1677 −19,100 −21,867 −6399 −121,293 −11.85 CO2 −175 −7406 −950 −20,675 −19 −29 −3 −1425 −1987 −841 −33,508 −16.46 Italy GDP −374 −27,077 −5882 −30,881 −349 −2085 −3379 −9570 −6776 −3449 −89,821 −12.10 CO2 −113 −8897 −461 −19,319 −2 −16 −8 −319 −490 −214 −29,838 −18.18 Cyprus GDP −1 −52 −65 −387 24 −6 16 −22 23 −46 −516 −5.62 CO2 0 −79 −6 −161 0 0 0 −2 1 −3 −249 −11.62 Latvia GDP −5 −63 32 −351 −44 −3 −18 −13 4 −90 −549 −5.00 CO2 −4 −32 5 −473 0 0 −1 −1 1 −8 −513 −9.01 Lithuania GDP 6 −35 5 −224 55 −1 36 −24 −16 −43 −241 −1.35 CO2 3 −26 0 −598 1 0 0 −1 −2 −4 −627 −3.22 Luxembourg GDP −1 −226 −198 −543 244 −37 43 −23 118 −7 −629 −2.45 CO2 −1 −121 −11 −1090 2 −1 0 −1 10 −1 −1213 −12.78 Hungary GDP −24 −1351 −153 −723 120 83 −73 −330 −648 −201 −3299 −6.45 CO2 −13 −726 −34 −671 7 4 −4 −67 −134 −31 −1669 −7.97 Netherlands GDP 105 −1795 114 −6282 −214 203 442 −2965 −3751 −1706 −15,850 −4.86 CO2 81 −1019 10 −6345 −1 2 3 −197 −253 −257 −7977 −7.28 Austria GDP −69 −3803 −351 −6278 168 −7 391 −2648 21 −1020 −13,595 −8.55 CO2 −23 −1531 −43 −2640 1 0 0 −70 1 −65 −4370 −13.34 Poland GDP −169 −3146 163 −4293 171 47 138 −29 1012 −1107 −7212 −3.58 CO2 −386 −2507 3 −6755 5 11 7 −6 308 −279 −9600 −6.36 Portugal GDP −87 −1763 97 −3542 51 −130 82 −991 −417 −484 −7183 −8.73 CO2 −50 −1074 19 −3453 0 −1 0 −55 −43 −52 −4708 −14.38 Romania GDP −131 −3241 503 −760 678 −41 7 −38 151 −787 −3659 −5.33 CO2 −23 −1932 186 −620 18 −3 0 −5 32 −199 −2546 −9.03 Slovenia GDP −33 −412 −46 −518 −24 10 13 −186 −61 −89 −1347 −7.22 CO2 −11 −157 −28 −345 −1 0 0 −47 −8 −23 −621 −9.66 Slovakia GDP −28 −1363 −424 −707 −81 −125 107 −230 25 −141 −2968 −8.05 CO2 −3 −1144 −150 −687 −1 −2 4 −32 4 −51 −2062 −11.71 Finland GDP −53 −773 −20 −1471 96 133 −35 −347 −363 −484 −3315 −3.48 CO2 −18 −283 −2 −1714 0 5 0 −15 −25 −44 −2097 −7.39 Sweden GDP −5 −3549 169 −3294 241 324 294 −1417 −787 −193 −8214 −3.81 CO2 −2 −868 15 −2281 1 1 2 −62 −27 −10 −3230 −7.96 Total GDP −584 −129,182 −21,540 −155,549 −3713 −4588 −5224 −62,621 −48,111 −30,209 −461,320 −8.51 CO2 −338 −49,562 −1343 −133,838 −27 −23 −3 −3214 −4068 −3194 −195,610 −12.11 Table A3 Percentage contribution of economic activities to GDP and CO2 variation. Table A3 A C F G, H, I J K L M, N O, P, Q R, S, T, U Belgium GDP 0.36 −12.50 −6.30 −43.14 −1.87 −1.81 1.34 −15.66 −15.34 −5.09 CO2 1.02 −19.31 −1.41 −68.78 −0.09 −0.03 0.01 −3.24 −4.16 −4.01 Bulgaria GDP −1.67 −49.41 −3.50 −59.80 10.47 5.77 −19.71 4.05 20.07 −6.26 CO2 −0.49 −28.81 −0.39 −71.18 0.05 0.00 −0.03 0.16 0.78 −0.09 Czech Rep. GDP 1.37 −50.52 −1.26 −39.11 1.83 −2.00 −3.26 −3.90 0.34 −3.49 CO2 1.15 −31.73 −0.28 −67.76 0.03 −0.02 −0.03 −1.14 0.03 −0.25 Denmark GDP 2.35 −11.23 −5.37 −50.32 −0.16 5.71 −1.01 1.33 −19.74 −21.54 CO2 1.26 −1.18 −0.41 −98.42 0.00 0.02 0.00 0.04 −0.70 −0.60 Germany GDP −0.06 −48.17 2.66 −17.45 −1.10 0.28 −0.02 −16.10 −13.40 −6.64 CO2 −0.10 −42.33 0.91 −50.18 −0.06 0.01 0.00 −1.95 −4.11 −2.19 Estonia GDP 2.87 −59.85 38.14 −91.14 38.98 9.66 −25.13 −9.38 7.34 −11.49 CO2 0.52 −21.15 1.96 −80.38 0.15 0.05 −0.66 −0.49 1.46 −1.46 Spain GDP 0.90 −17.31 −8.46 −50.49 −2.48 0.14 −2.06 −11.66 0.52 −9.09 CO2 0.57 −15.42 −0.09 −84.44 −0.04 0.00 0.00 −0.25 0.06 −0.39 France GDP −0.39 −16.44 −9.26 −28.95 −1.84 −2.69 −1.38 −15.75 −18.03 −5.28 CO2 −0.52 −22.10 −2.83 −61.70 −0.06 −0.09 −0.01 −4.25 −5.93 −2.51 Italy GDP −0.42 −30.15 −6.55 −34.38 −0.39 −2.32 −3.76 −10.65 −7.54 −3.84 CO2 −0.38 −29.82 −1.55 −64.75 −0.01 −0.05 −0.03 −1.07 −1.64 −0.72 Cyprus GDP −0.19 −10.05 −12.51 −75.01 4.59 −1.12 3.06 −4.26 4.40 −8.89 CO2 −0.09 −31.67 −2.35 −64.45 0.07 −0.02 0.01 −0.71 0.39 −1.20 Latvia GDP −0.89 −11.44 5.86 −63.87 −7.92 −0.53 −3.26 −2.35 0.71 −16.32 CO2 −0.70 −6.30 1.03 −92.05 −0.08 −0.01 −0.21 −0.27 0.10 −1.50 Lithuania GDP 2.57 −14.68 1.87 −93.07 22.65 −0.37 15.10 −9.79 −6.64 −17.63 CO2 0.41 −4.13 0.02 −95.28 0.09 0.00 0.04 −0.17 −0.28 −0.70 Luxembourg GDP −0.16 −35.83 −31.46 −86.29 38.74 −5.82 6.77 −3.64 18.78 −1.10 CO2 −0.06 −10.00 −0.91 −89.87 0.17 −0.05 0.03 −0.07 0.83 −0.07 Hungary GDP −0.74 −40.96 −4.62 −21.91 3.63 2.53 −2.20 −10.00 −19.63 −6.09 CO2 −0.81 −43.52 −2.06 −40.20 0.41 0.25 −0.21 −4.02 −8.01 −1.83 Netherlands GDP 0.66 −11.33 0.72 −39.63 −1.35 1.28 2.79 −18.71 −23.67 −10.76 CO2 1.02 −12.78 0.12 −79.55 −0.01 0.02 0.04 −2.47 −3.17 −3.22 Austria GDP −0.50 −27.97 −2.58 −46.18 1.24 −0.05 2.87 −19.48 0.16 −7.50 CO2 −0.52 −35.04 −0.98 −60.41 0.03 0.00 0.01 −1.61 0.02 −1.49 Poland GDP −2.35 −43.62 2.26 −59.52 2.38 0.65 1.91 −0.41 14.03 −15.35 CO2 −4.02 −26.12 0.03 −70.37 0.05 0.12 0.07 −0.06 3.21 −2.91 Portugal GDP −1.21 −24.54 1.35 −49.31 0.70 −1.81 1.14 −13.79 −5.80 −6.74 CO2 −1.06 −22.82 0.41 −73.33 0.01 −0.03 0.00 −1.18 −0.91 −1.10 Romania GDP −3.59 −88.58 13.74 −20.76 18.53 −1.13 0.20 −1.04 4.13 −21.50 CO2 −0.90 −75.90 7.31 −24.35 0.70 −0.11 0.01 −0.21 1.26 −7.82 Slovenia GDP −2.44 −30.59 −3.45 −38.45 −1.79 0.71 0.99 −13.84 −4.51 −6.62 CO2 −1.82 −25.37 −4.49 −55.64 −0.20 0.05 0.04 −7.55 −1.27 −3.75 Slovakia GDP −0.95 −45.92 −14.28 −23.81 −2.74 −4.22 3.60 −7.76 0.84 −4.76 CO2 −0.16 −55.46 −7.28 −33.31 −0.05 −0.08 0.18 −1.57 0.21 −2.48 Finland GDP −1.60 −23.31 −0.60 −44.37 2.90 4.02 −1.06 −10.47 −10.94 −14.59 CO2 −0.87 −13.49 −0.10 −81.75 0.00 0.25 −0.01 −0.74 −1.18 −2.12 Sweden GDP −0.06 −43.20 2.06 −40.10 2.94 3.95 3.58 −17.24 −9.58 −2.34 CO2 −0.05 −26.87 0.46 −70.61 0.02 0.04 0.05 −1.92 −0.83 −0.30 Total GDP −0.13 −28.00 −4.67 −33.72 −0.80 −0.99 −1.13 −13.57 −10.43 −6.55 CO2 −0.17 −25.34 −0.69 −68.42 −0.01 −0.01 −0.00 −1.64 −2.08 −1.63 1 NACE (Nomenclature statistique des activités économiques dans la Communauté européenne) is the statistical classification of economic activities in the European Community. ==== Refs References Arthi V. Parman J. Disease, downturns, and wellbeing: economic history and the long-run impacts of COVID-19 Explor. Econ. Hist. 2020 101381 (in press) Baldasano M.J. COVID-19 lockdown effects on air quality by NO2 in the cities of Barcelona and Madrid (Spain) Sci. Total Environ. 741 2020 140353 32593894 Bao R. Zhang A. Dees lockdown reduce air pollution? Evidence from 44 cities in northern China Sci. Total Environ. 731 2020 139052 32413655 Berman D. Ebisu K. Changes in U.S. air pollution during the COVID-19 pandemic Science of the Total Environment 739 2020 139864 Bulut M. Analysis of the Covid-19 impact on electricity consumption and production 2020 preprints.org 10.20944/preprints202010.0348.v1 Chauhan A. Singh R.P. Decline in PM2.5 concentrations over major cities around the world associated with COVID-19 Environmental Research 187 2020 109634 Cheshmehzangi A. COVID-19 and household energy implications: what are the main impacts on energy use? Heliyon 6 10 2020 e05202 Collivignarelli M.C. Abba A. Bertanza G. Ptdrazzani R. Ricciardi P. Miino C.C. Lockdown for COVID-19 in Milan: what are the effects on air quality? Sci. Total Environ. 732 2020 139280 32402928 Dantas G. Siciliano B. Franca B.B. da Silva C.M. Arbilla G. The impact of COVID-19 partial lockdown on the air quality of the city of Rio de Janeiro, Brazil Sci. Total Environ. 729 2020 139085 32361428 EC Policy measures taken against the spread and impact of the coronavirus – 20 August 2020 Available at: https://ec.europa.eu/info/live-work-travel-eu/health/coronavirus-response/jobs-and-economy-during-coronavirus-pandemic_en 2020 ECDC – European Center for Disease Prevention and Control Data on country responses measure Available at: https://www.ecdc.europa.eu/en/publications-data/download-data-response-measures-covid-19 Ferrarini A. Barbieri E. Biggeri M. Di Tommaso R. Industrial policy for sustainable human development in the post-COVID19 era World Dev. 137 2021 105215 32994664 Friedlingstein P. Global carbon budget 2019 Earth Syst. Sci. Data 11 4 2019 1783 1838 Han E. Tan M.M.J. Turk E. Sridhar D. Leung G.M. Lessons learnt from easing COVID-19 restrictions: an analysis of countries and regions in Asia Pacific and Europe Lancet 2020 10.1016/S0140-6736(20)32007-9 (Accessed: 12 November 2020) Han P. Cai Q. Oda T. Zeng N. Shan Y. Lin X. Liu D. Assessing the recent impact of COVID-19 on carbon emissions from China using domestic economic data Sci. Total Environ. 750 2021 141688 32835964 He W. Zhang J. Li W. Information technology solutions, challenges, and suggestions for tackling the COVID-19 pandemic International Journal of Information Management 2020 102287 (in press) Carbon Monitor website Available at: http://carbonmonitor.org WHO Covid-19 dashboard Available at: https://covid19.who.int/ EC website – European Climate Change Programme Available at: https://ec.europa.eu/clima/policies/eccp_en Eurostat Air Emission database - Air Emission Intensities by NACE Rev. 2 Activity Available at: https://ec.europa.eu/eurostat/databrowser/view/env_ac_aeint_r2/default/table?lang=en Eurostat construction database Available at: https://ec.europa.eu/eurostat/databrowser/view/sts_copr_m/default/table?lang=en Eurostat industrial production database Available at: https://ec.europa.eu/eurostat/databrowser/view/sts_inpr_m/default/table?lang=en Eurostat services database Available at: https://ec.europa.eu/eurostat/databrowser/view/sts_setu_m/default/table?lang=en Eurostat retail trade database Available at: https://ec.europa.eu/eurostat/databrowser/view/sts_trtu_m/default/table?lang=en Eurostat website – COVID-19 statistics Available at: https://ec.europa.eu/eurostat/web/covid-19/data EEA website https://www.eea.europa.eu/highlights/sharpest-decrease-of-the-decade International Monetary Fund Website – Policy responses to COVID-19 Available at: https://www.imf.org/en/Topics/imf-and-covid19/Policy-Responses-to-COVID-19 World Health Organization website: https://www.who.int/emergencies/diseases/novel-coronavirus-2019 Ibn-Mohammed T. Mustapha K.B. Godsell J. Adamu Z. Babatunde K.A. Akintade D.D. Acquaye A. Fujii H. Ndiay M.M. Yamoah F.A. Koh S.C.L. A critical analysis of the impacts of COVID-19 on the global economy and ecosystems and opportunities for circular economy strategies Resour. Conserv. Recycl. 164 2020 105169 32982059 International Energy Agency Global Energy Review 2020: the impacts of the COVID-19 crisis on global energy demand and CO2 emissions Available at: https://www.iea.org/reports/global-energy-review-2020 2020 Isaifan R.J. The dramatic impact of Coronavirus outbreak on air quality: has it saved as much as it has killed so far? Glob. J. Environ. Sci. Manag. 6 2020 275 288 Jaspal R. Lopes B. Lopes P. Predicting social distancing and compulsive buying behaviors in response to COVID-19 in a United Kingdom sample Cogent Psychol. 6 2020 1800924 Klemes J.J. Van Fan Y. Jiang P. The energy and environmental footprints of COVID-19 fighting measures – PPE, disinfection, supply chains Energy 15 221 2020 118701 Le Quere C. Jackson R.B. Jones M.W. Temporary reduction in daily global CO2 emissions during the COVID-19 forced confinement Nat. Clim. Chang. 10 2020 674 683 Li L. Li Q. Huang L. Wang Q. Zhu A. Xu J. Air quality changes during the COVID-19 lockdown over the Yangtze River Delta region: an insight into the impact of human activity pattern changes on air pollution variation Sci. Total Environ. 732 2020 139282 32413621 Liu Z. Near-real-time monitoring of global CO2 emissions reveals the effects of the COVID-19 pandemic Nat. Commun. 11 2020 5172 33057164 Mahato S. Pal S. Ghosh K.G. Effect of lockdown amid COVI-19 pandemic on air quality of the megacity Delhi, India Sci. Total Environ. 730 2020 139076 Manzanedo R.D. Manning P. COVID-19: lessons for the climate change emergency Sci. Total Environ. 742 2020 140563 32619845 Menut L. Bessagnet B. Siour G. Mailler S. Pennel R. Cholakian A. Impact of lockdown measures to combat Covid-19 on air quality over western Europe Sci. Total Environ. 741 2020 140426 32593893 Myllyvirta L. Analysis: Coronavirus temporarily reduced China's CO2 emissions by a quarter Available at: https://www.carbonbrief.org/analysis-coronavirus-has-temporarily-reduced-chinas-co2-emissions-by-a-quarter 2020 Nakada L.Y.L. Urban R.C. COVID-19 pandemic: impacts on the air quality during the partial lockdown in Sao Paulo state, Brazil Sci. Total Environ. 730 2020 139087 32380370 OECD Preliminary Report: Evaluation of the Impact of the Coronavirus (COVID-19) on Fruits and Vegetable Trade. TAD/CA/FVS/WD(2020)1/REV7 2020 OECD Housing amid Covid-19: policy responses and challenges Available at: https://read.oecd-ilibrary.org/view/?ref=135_135322-lju8yj5sib&title=Housing-Amid-Covid-19-Policy-Responses-and-Challenges 2020 Ordonez C. Garrido-Perez J.M. Herrera-Garcia R. Early spring near-surface ozone in Europe during the COVID-19 shutdown: meteorological effects outweigh emission changes Sci. Total Environ. 747 2020 141322 32781318 Rugani B. Caro D. Impact of Covid-19 outbreak measures of lockdown on the Italian Carbon Footprint Sci. Total Environ. 737 2020 139806 32492608 Rume T. Islam S.M.D.-U. Environmental effects of COVID-19 pandemic and potential strategies of sustainability Heliyon 6 2020 c04965 Sarkodie S.A. Owusu P.A. Global assessment of environment, health and economic impact of the novel coronavirus (COVID-19) Environment, Development and Sustainability 2020 10.1007/s10668-020-00801-2 (accessed: 12 November 2020) Shakil M.H. Munim Z.H. Tasnia M. Sarowar S. COVID-19 and the environment: a critical review and research agenda Sci. Total Environ. 745 2020 141022 32711074 Sharma T. Jain S. Social and travel lockdown impact considering coronavirus disease (COVID-19) on air quality in megacities of India: present benefits, future challenges and way forward Aerosol Air Qual. Res. 20 2020 1222 1236 Sharma S. Zhang M. Gao J. Zhang H. Kota S.H. Effect of restricted emissions during COVID-19 on air quality in India Sci. Total Environ. 728 2020 138878 32335409 Sicard P. De Marco A. Agathokleous E. Feng Z. Xu X. Paoletti E. Amplified ozone pollution in cities during the COVID-19 lockdown Sci. Total Environ. 735 2020 139542 32447070 Sovacool B.K. Del Rio D.F. Griffiths S. Contextualizing the Covid-19 pandemic for a carbon-constrained world: insights for sustainability transitions, energy justice, and research methodology Energy Res. Soc. Sci. 68 2020 101701 32844087 Stoll C. Mehling M.A. COVID-19: clinching the climate opportunity One Earth 3 4 2020 400 404 34173539 Tobia A. Carnerero C. Reche C. Massague J. Via M. Minguillon M.C. Changes in air quality during the lockdown in Barcelona (Spain) one month into the SARS-CoV-2 epidemic Sci. Total Environ. 726 2020 138540 32302810 Wang H. Lu X. Deng Y. Sun Y. Nielsen C.P. Liu Y. Zhu G. Bu M. Bi J. McElroy M.M. China’s CO2 peak before 2030 implied from characteristics and growth of cities Nat. Sustain. 2 8 2019 748 754 World Bank COVID-19 to plunge global economy into worst recession since World War II Available at: https://www.worldbank.org/en/news/press-release/2020/06/08/covid-19-to-plunge-global-economy-into-worst-recession-since-world-war-ii 2020 World Bank Pandemic, recession: the global economy in crisis Available at: https://www.worldbank.org/en/publication/global-economic-prospects 2020 World Bank Unmasking the impact of COVID-19 on businesses - firm level evidence from across the world Policy Research Working Paper 9434 2020 Zambrano-Monserrate M.A. Ruano M.A. Alcade L.S. Indirect effects of COVID-19 on the environment Sci. Total Environ. 278 2020 138813

33477052

PMC9754059

NO-CC CODE

2022-12-16 23:26:25

no

Sci Total Environ. 2021 May 15; 769:145115

utf-8

Sci Total Environ

10.1016/j.scitotenv.2021.145115

oa_other

==== Front Soc Sci Res Soc Sci Res Social Science Research 0049-089X 1096-0317 The Authors. Published by Elsevier Inc. S0049-089X(21)00169-1 10.1016/j.ssresearch.2021.102692 102692 Article Belief change in times of crisis: Providing facts about COVID-19-induced inequalities closes the partisan divide but fuels intra-partisan polarization about inequality Mijs Jonathan J.B. ab∗ de Koster Willem b van der Waal Jeroen b a Department of Sociology, Boston University, 100 Cummington Mall, Boston, MA, 02215, United States b Department of Public Administration and Sociology, Erasmus University Rotterdam, PO Box 1738, DR Rotterdam, 3000, the Netherlands ∗ Corresponding author 21 12 2021 5 2022 21 12 2021 104 102692102692 11 5 2021 4 11 2021 18 12 2021 © 2021 The Authors 2021 Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. Population-based survey research demonstrates that growing economic divides in Western countries have not gone together with increased popular concern about inequality. Extant explanations focus on ‘misperception’: people generally underestimate the extent of inequality and overestimate society's meritocratic nature. However, scholarly attempts to correct people's misperceptions have produced mixed results. We ask whether COVID-19, by upending everyday life, has made people responsive to information about inequality, even if that entails crossing ideological divides. We field an original survey experiment in the United States, a least-likely case of belief change, given high levels of inequality and partisan polarization. Our informational treatment increases (1) concerns over economic inequality, (2) support for redistribution, and (3) acknowledgement that COVID-19 has especially hurt the most vulnerable. Information provision renders non-significant the partisan gap between moderate Democrats and Republicans but increases that between moderate and strong Republicans. We discuss our findings' implications and suggestions for further research. Keywords COVID-19 Inequality Information provision Political polarization Support for redistribution ==== Body pmc1 Introduction The steep rise in economic inequality starting in the 1970s presents a puzzle to scholars of public opinion: while some studies describe growing support for government intervention (Kenworthy and Pontusson, 2005), in many countries we see surprisingly little attitudinal evidence of concern or calls for income redistribution among the public at large (Luebker, 2014; Trump, 2017; Bradley et al., 2003; Loveless and Whitefield, 2011; Breznau and Hommerich, 2019). Similarly, while some research documents a growing sense of injustice among certain segments of society (for the case of Germany, see: Sachweh and Sthamer, 2019), large-scale survey research reports stable or strengthening popular belief in meritocracy (Kelly and Enns, 2010; Kenworthy and McCall, 2008; Mijs, 2018, Mijs, 2021; Suhay et al., 2020). These patterns are typically explained by widespread ‘misperception’ of the nature and extent of economic disparities: many people confidently hold incorrect convictions and perceptions, which keeps them from seeing the full extent and non-meritocratic nature of economic inequality (Jerit and Zhao, 2020; Kuklinski et al., 2000; McCall et al., 2017). Whereas misperceptions are endemic, addressing them through the provision of factual information is complicated by ideological camps' distrust in science, experts and government (Pechar et al., 2018), different exposure to news and information (Flaxman et al., 2016; Mummolo, 2016), and, crucially, by partisan motivated reasoning in a polarized political landscape: “individuals interpret information through the lens of their party commitment” (Bolsen et al., 2014, p. 235). Evidence for attitudinal effects of informational interventions is mixed. While learning about high levels of inequality generally raises concerns about inequality (McCall et al., 2017), it typically generates support for redistribution only among people already positively inclined and trusting in government (Alesina et al., 2018; Kuziemko et al., 2015). Concerns over economic disparities might hardly be affected by COVID-19-induced surges in inequality, as the pandemic continues to split the public along partisan lines (Druckman et al., 2021; Grasso et al., 2021; Kreps and Kriner, 2020; Shepherd et al., 2020). However, research on past crises suggests that such “unsettled times” (Swidler, 1986) may inspire belief change (Bisgaard, 2015; Gidron and Mijs, 2019; Lee and Fujita, 2011; Margalit, 2013; Naumann et al., 2016). This paper investigates belief change in times of the COVID-19 crisis. Specifically, we ask whether information on COVID-19-induced inequalities heightens the public's concern about inequity and strengthens support for government intervention, and how the uptake of information is shaped by partisanship. Following fruitful prior empirical applications (e.g. Neimanns et al., 2018), we leverage a survey experiment methodology to analyze the effects, if any, of factual information describing the impact of the pandemic on people's beliefs about inequality, COVID-19, and the role of government. We field our study in the United States, which we consider a strategic least likely case (Gerring, 2007) of belief change. The US stands out from other advanced economies because of its high levels of economic inequality, coupled with relatively limited public support for income redistribution and deep political polarization, as demonstrated by extant research utilizing population-based surveys (Alesina et al., 2018; Atkinson et al., 2011; Kozlowski and Murphy, 2021). Hence, by fielding our study in the US, we aim to make a conservative assessment of the potential of informational interventions. Our results suggest that learning about COVID-19-induced inequalities inspires (1) stronger acknowledgement that COVID-19 has especially hurt the most economically vulnerable, (2) more concerns over economic inequality, and, crucially, (3) stronger support for income redistribution. Importantly, conditional treatment effects by partisanship reveal larger effects among moderate Republicans. Consequently, our informational treatment renders non-significant the partisan gap between moderate Democrats and Republicans, while the gap between moderate and strong Republicans widens. Thus, our findings suggest that uniform information on a crisis can increase concerns about inequality among substantial parts of the population. It can even overcome attitudinal gaps between moderate partisans in a country renowned for widespread polarization and relatively high tolerance of inequality. In what follows we situate our study in the broader literature on crises and political polarization before discussing our research design and results. We conclude by discussing how our findings speak to other country contexts and provide suggestions for further research. 2 Belief change in times of crisis Past research documents how tumultuous times can cause a shift in a conservative (Schüller, 2015) or liberal direction (Eadeh and Chang, 2020). For instance, scholars have documented how domestic and international crises, from the Ebola outbreak in the US (Beall et al., 2016) to terrorist attacks on American soil (Schüller, 2015) and abroad (Berrebi and Klor, 2008), have shifted public opinion in a conservative direction. Such a conservative response to systemic threat arguably reflects a heightened need to protect the status quo—no matter its unequal nature (Jost et al., 2003). Recent surveys suggest that a similar pattern may be unfolding today: Cappelen et al. (2020), in an unpublished survey of 8,000 Americans reported in The New York Times, find that people who were primed to think about the impact of the virus on their community cared more about their country and cared less about inequality. Other research suggests that the very scope of a crisis like the COVID-19 pandemic may make it a ‘common enemy,’ which may instead increase support for government intervention rather than retrenchment (cf. Oude Groeniger et al., 2021). Eadeh and Chang (2020) suggest that a crisis can cause a 'liberal turn' if the threat concerns policy areas like health care and the environment, where liberal politicians are perceived to be more competent. The New Deal, for instance, can be seen as a direct response to the Great Depression (Gordon, 2016) and, more recently, the Great Recession led the Obama administration to the greatest redistributive effort in three decades (CBO, 2019). In this particular historical moment, public appreciation of ‘essential workers’ may go together with changing perceptions, attitudes and policies addressing inequality (Waterfield, 2020). President Biden's Build Back Better Framework, arguably, is a step in that direction. At the same time, it is unlikely that Americans across the political divide have experienced and understood the pandemic in the same way. Such is the finding of a Californian survey on the perceived impact of the COVID-19 pandemic on economic inequality: over 80 percent of Democrats but just 40 percent of Republicans believed economic inequality had increased (Mora et al., 2020; Shepherd et al., 2020). A similar partisan divide characterizes Americans’ concerns about the coronavirus more generally—and the gap is widening, as suggested by weekly surveys between early and late 2020 (Civiqs, 2020).Hypothesis 1 Americans’ beliefs about COVID-19, economic inequality and the role of government are polarized along party lines. Compounding the ideological divide, the public's political response to COVID-19 likely depends on how the crisis is understood by different ideological camps (Bird and Ritter, 2020; Cox, 2001; Tierney, 2007). Theorizing on partisan motivated reasoning (Bolsen et al., 2014) suggests that information about COVID-19-induced inequalities is filtered through a partisan lens, making some parts of the population more receptive to it than others. Republicans are politically motivated to accept economic inequalities as deserved, to oppose government intervention in the economy, and to be less supportive and compliant with COVID-19-related government interventions more specifically (cf. Conway et al., 2021; Gollwitzer et al., 2020). Conversely, Democrats traditionally express more concerns about inequality, and are more supportive of income redistribution (Kozlowski and Murphy, 2021; Pechar et al., 2018) and of public spending on COVID-19-related healthcare (Gollwitzer et al., 2020). It follows that provision of the same factual information may have a different impact across partisan lines, leaving Republicans unaffected while resonating with Democrats who may already be so inclined (Grossman et al., 2020).Hypothesis 2A Information about COVID-19-induced inequalities heightens concerns about inequality and strengthens support for government intervention only among Democrats. An alternative perspective considers a crisis, more fundamentally, as a “plastic hour” (Gershom, cited in Packer, 2020, p. 50) in which taken-for-granted practices, policies, and attitudes are upended. Could such 'unsettled times' (Swidler, 1986) constitute an 'event' that can shock or rupture political divides (Wagner-Pacifici, 2010)? Specifically, may times like these make people responsive to information about inequality, even if doing so means crossing ideological rifts? Research from the United Kingdom suggests the 2008 Great Recession did just that. Bisgaard (2015, p. 840) describes a pre-crisis partisan gap in perceptions about the economy which “evaporates” during the crisis, as even the staunchest partisans share the dire diagnosis of their country's economic state, even if they disagree about where to lay blame. Research on the Netherlands (Gidron and Mijs, 2019), Germany (Naumann et al., 2016), and the United States (Margalit, 2013) similarly describes growing support for redistribution across the political spectrum in times of crisis. Given its deep impact across society as a combined economic and public health crisis, the COVID-19 pandemic arguably constitutes a greater ‘rupture’ in everyday life than most other crises: scholars observe a “tsunami of change” and note that “the unusual conditions of the pandemic – unlike other crises – have impacted almost every facet of our lives” (Robinson et al., 2021, pp. 1608–9). As such, it may produce a particular ‘plastic’ moment.Hypothesis 2B Information about COVID-19-induced inequalities heightens concerns about inequality and strengthens support for government intervention on both sides of the political divide. 3 Data and Methods 3.1 Survey design Previous studies suggest that the most effective informational treatments are designed as non-partisan, cognitively light (Alesina et al., 2018), informational interventions designed to ‘shock’ participants' belief system (Kuziemko et al., 2015). Incorporating these insights, we developed an ‘omnibus treatment’ (cf. Kuziemko et al., 2015) describing COVID-19's economic consequences (Supplementary Information, Fig. S1). Participants were shown a graph of the number of Americans filing for unemployment between January 2020 and July 2020. The graph is accompanied by facts taken from various trusted, nonpartisan, sources, which (1) highlight the total number of people who filed for unemployment (cf. Bureau of Labor Statistics, 2020), (2) introduce the prognosis that this crisis will have a larger economic effect than any other crisis in recent history (cf. Cox, 2020; DeRensis, 2020; Schwartz, 2020), (3) emphasize its disproportionate effects on low and middle-income workers (cf. Athreya et al., 2020), and (4) inform participants that, meanwhile, some of the wealthiest Americans' fortunes have significantly grown (cf. Collins, 2021). Participants in the control condition were presented an unrelated but similarly looking graph depicting what share of different age groups are getting enough exercise, accompanied with facts about the positive health effects of physical exercise and stating the share of youth and adults that meets the recommended level of sports and exercise (SI, Fig. S2). The treatment was embedded in a between-subject survey design incorporating pretreatment and post-treatment questions. As in a standard between-subject design, we identify the treatment effect as the difference in post-treatment responses between participants in the treatment and control condition. Incorporating pretreatment questions that are distinct from but correlated with our post-treatment questions produces higher precision and more statistical power than a standard between-subjects design (Clifford et al., 2021; Lin, 2013). Specifically, we asked three questions which are correlated with the post-treatment questions about inequality, COVID-19 and the role of government (0.21≤ r ≤ 0.60) and include these as pretreatment controls in regression models estimating the treatment effect. This means that participants in both the control and treatment condition are introduced to the topic of inequality prior to our measurement of their post-treatment beliefs. As such, our design produces a conservative estimate of the effect of information over and above a baseline level of inequality priming. 3.2 Data We set to recruit 1,000 participants using a quota sample provided by Prolific Academic stratified by sex, age and race/ethnicity to match US Census Current Population Statistics. We recruited participants between August 5 and August 11, 2020, through Prolific Academic. Prolific is a survey firm specializing in social science research, founded by academics in Oxford, UK. It has worked with researchers at top institutions around the world and compares favorably to other survey firms that offer high-quality alternatives to Amazon Mechanical Turk (Palan and Schitter, 2018). We fielded our survey experiment with Prolific's active panel of 138,363 participants based in the US. Panelists are registered after verification of a valid e-mail address, phone number and payment method. Each panelist is assigned a unique identifier, matched with self-reported basic demographic information. They receive compensation for each survey completed, after an evaluation of their survey responses. Panelists flagged for low-quality responses more than once are removed from the panel. We obtained a sample of 1,003 participants. Ten participants (one percent) did not complete the survey. New participants were recruited in their place. The final sample matches population statistics on race and gender but skews slightly toward a younger demographic (SI, Table S3). Participants were randomly assigned to either the control (n = 500) or treatment condition (n = 503). Based on power calculations, we ensured that treatment and control group had 500 participants per condition to get a power of 0.9 when the Cohen's d = 0.2. We obtain almost perfect post-allocation balance between participants in the control and treatment group on key dimensions (SI, Table S4). We took several measures to secure the quality of our research. First, to accommodate people differently affected by COVID-19, working and not working, with and without caring duties, we provided an extended window, spanning two working days and a weekend day, during which participants could take the survey. Second, we designed the survey to be short: the median time of completion was 11 min. Third, we tested our questions and treatment design in two pilot surveys (n = 100 and n = 150). Fourth, to minimize selection bias, we gave our survey a non-descript name (“Social topics in the United States”) and set compensation at a relatively generous $2.50, corresponding to an hourly rate of approximately $14. Fifth, we include a post-treatment attention check, by asking participants which informational treatment they were given (coronavirus, exercise, dining, don't know). Only eighteen participants (1.8 percent) failed the check, which indicates that respondents were generally attentive. Those who failed the check were kept in the analysis so as not to induce bias (Aronow et al., 2019). The main results are qualitatively equivalent to including an attention check dummy (Table 2; Check B). Finally, we ran checks for survey straightlining, but found no concerning patterns in our data. We have made our data and code publicly available on the Open Science Foundation platform. 3.3 Measures We focus on four attitudinal variables about inequality, COVID-19, and the role of government, measured on a 7-point scale ranging from “Strongly disagree” to “Strongly agree”: 1) “differences in income in the United States are too large,” 2) “it is the responsibility of the government to reduce the differences in income between people with high incomes and those with low incomes,” 3) “the most vulnerable in society are hit hardest by the coronavirus, also known as COVID-19,” and 4) “the measures taken against the coronavirus are more harmful than the virus itself.” Questions 1 and 2 are adopted from the International Social Survey Programme Social Inequality module to allow for a direct comparison with international research (ISSP Research Group, 2018). Question 3 and 4 are original questions designed to capture strong sentiments about the impact of the pandemic as expressed on different sides of the political aisle: whereas the former taps into to concerns about COVID-19-related inequality most frequently found among Democrats (Mora et al., 2020; Shepherd et al., 2020), the latter reflects a longstanding concern of economic conservatives regarding the unemployment-inducing nature of government interference in the economy, more typically found among Republicans (Bruine de Bruin et al., 2020; Van der Waal et al., 2007). Because of indications of substantial intra-party variation in COVID-19-related attitudes (e.g., Havey, 2020), we use a more fine-grained measure of partisanship than most COVID-19 studies to date. To concisely discern weak from strong partisans, we asked for participants’ self-placement on a 10-point scale ranging from strong Democrat to strong Republican, using the middle as a default starting position (cf. Dalton, 2008). In the analyses below, we compare participants strongly identifying as Democrat (0–1) to those identifying as Democrat (2–3), those identifying as Republican (7–8), strongly identifying as Republican (9–10) and those in the middle (4–6). Fifty participants (5 percent) opted out of the question and were grouped with the middle category. Excluding these participants, in a robustness check, does not change our findings (Table 2; Check C). As a final robustness check, we replicate our main analyses by including additional pretreatment controls for age, gender, education, parental education, marital status, household income, employment status, self-placement on the social ladder, financial assets, religion and the date the survey was taken (for descriptive statistics, see SI, Table S5). Doing so, we obtained qualitatively equivalent results (Table 2; Check A). Table 1 provides descriptives for key dependent and independent variables.Table 1 Sample descriptives (n = 1,003). Table 1Variable Mean SD Dependent variables Differences in income too large (1–7) 5.74 1.37 Government responsibility to reduce differences in income (1–7) 4.98 1.80 Most vulnerable are hit hardest by COVID-19 (1–7) 6.11 1.14 Measures taken against COVID-19 are more harmful than the virus (1–7) 2.82 1.91 Independent variables Treatment assignment (0/1) 0.50 Party identification  Strong Democrat 0.30  Democrat 0.19  Neither 0.33  Republican 0.11  Strong republican 0.08 Pretreatment controls Society is fair when hard-working people earn more (1–7) 4.86 1.45 Racial diversity makes America stronger (1–7) 5.75 1.48 For society to be fair, income differences should be small (1–7) 4.87 1.61 4 Results 4.1 Ideological divide in beliefs about inequality, COVID-19 and role of government To address Hypothesis 1 regarding the political polarization of beliefs about inequality, COVID-19 and the role of government, we first present histograms to visualize the distribution of responses by party identification. Fig. 1 a describes a monotonic partisan divide, both in concerns about income inequality and support for redistribution, which corroborates Hypothesis 1. Participants identifying as strong Democrats are consistently more concerned about inequality than participants in the middle, and, in turn, than Republicans and strong Republicans (6.4 > 6.1 > 5.5 > 5.0 > 4.4; significantly different at p < .05), and more supportive of income redistribution (6.0 > 5.5 > 4.4 > 4.0 ≥ 3.6; all but the latter two are significantly different at p < .05).Fig. 1 Histogram of beliefs about (a) inequality and (b) COVID-19 by party identification. Note. Bars indicate the percentage of responses across response categories within each group. Question wording: “Differences in income in the United States are too large”; “It is the responsibility of the government to reduce the differences in income between people with high incomes and those with low incomes”; “The most vulnerable in society are hit hardest by the coronavirus (also known as COVID-19)”; and “The measures taken against the coronavirus are more harmful than the virus itself.” Fig. 1 Fig. 1b shows that strong Democrats and Democrats are also monotonically more convinced than Republicans and participants in the middle that the coronavirus has disproportionally affected vulnerable populations (6.5 > 6.3 > 5.9 > 5.5 ≥ 5.5; all contrasts but the latter two are significantly different at p < .05). Conversely, (strong) Republicans are more likely than (strong) Democrats to think that the measures taken to combat the virus have been more harmful than the virus itself, but Republicans not significantly more so than participants in the middle (2.0 < 2.4 < 3.2 ≤ 3.5 < 4.6). Taken together, our results describe a partisan divide both in perceptions of inequality and the coronavirus and in attitudes about the role the government, in line with Hypothesis 1. Below, we first consider the general effect of provision of information, before testing our Hypotheses regarding how partisanship shapes the effect of information. 4.2 Treatment effect of information describing COVID-19-induced inequalities To evaluate the effect of the informational treatment on participants’ beliefs about inequality, COVID-19 and the role of government, we identify the average treatment effect of information by estimating OLS regressions, modeling each outcome as a function of the information treatment, and the pretreatment controls described above (Table 2). Fig. 2 plots average marginal effects calculated from the regression models and shows a positive treatment effect for three out of four attitudes. The informational treatment is associated with a 0.26-point (95% CI, 0.12–0.40) increase in participants' belief that income inequality is too high, a 0.19-point (95% CI, 0.01–0.36) increase in support for government redistribution, and a 0.20-point (95% CI, 0.07–0.33) increase in the belief that COVID-19 has disproportionately affected society's most vulnerable groups. We do not find a significant treatment effect (p < .05) for participants' belief that government measures are worse than the virus.Fig. 2 Treatment effect on beliefs about inequality, COVID-19, and the role of government. Note. Average marginal effects are estimated from OLS regression models including pretreatment attitudes and all control variables. Whiskers indicate the 95% confidence interval around each estimate. Fig. 2 All in all, even in a strongly polarized country known for its high level of economic inequality and comparatively low levels of public concern about this, we find a substantively meaningful effect of information provision across three dimensions of inequality, between a fifth and a quarter of a point on a 7-point attitudinal scale. But how does factual information about COVID-19-induced inequalities affect participants’ beliefs across the political divide? 4.3 Treatment effect by party identification To directly assess Hypotheses 2A and 2B, Table 2 reports average treatment effects and conditional treatment effects for the main models and three robustness checks (see Data and Methods). Conditional effects are estimated by interacting participants’ treatment condition and party identification.Table 2 Results from main model and alternative specifications. Table 2 Income Redistribution Vulnerable Measures As reported in manuscript (n = 1,003) Average treatment effect 0.26 *** 0.19 * 0.20 ** 0.15 Conditional treatment effects  Strong democrat 0.27 * 0.21 0.22 −0.13  Democrat 0.18 0.04 0.23 0.17  Neither 0.26 * 0.18 0.10 0.24  Republican 0.54 ** 0.72 ** 0.43 * 0.06  Strong republican 0.03 −0.38 0.11 0.87 * Check A (n = 1,003) Income Redistribution Vulnerable Measures Average treatment effect 0.28 *** 0.22 ** 0.21 ** 0.12 Conditional treatment effects  Strong democrat 0.29 * 0.32 * 0.24 * −0.09  Democrat 0.20 0.11 0.21 0.16  Neither 0.30 * 0.12 0.10 0.19  Republican 0.51 * 0.72 ** 0.44 * 0.03  Strong republican 0.06 −0.23 0.12 0.74 Check B (n = 1,003) Income Redistribution Vulnerable Measures Average treatment effect 0.28 *** 0.23 ** 0.21 ** 0.13 Conditional treatment effects  Strong democrat 0.29 * 0.32 * 0.24 * −0.09  Democrat 0.20 0.11 0.21 0.15  Neither 0.30 * 0.12 0.10 0.19  Republican 0.51 * 0.71 ** 0.44 * 0.03  Strong republican 0.07 −0.15 0.16 0.79 Check C (n = 937) Income Redistribution Vulnerable Measures Conditional treatment effects  Strong democrat 0.29 * 0.32 * 0.24 * −0.08  Democrat 0.19 0.10 0.22 0.16  Neither 0.36 ** 0.12 0.06 0.42 *  Republican 0.54 ** 0.71 ** 0.45 * 0.03  Strong republican 0.03 −0.27 0.10 0.77 Note. ‘Check A’ reports estimation results from OLS models including a bank of control variables; ‘Check B’ adds an attention check dummy to the controls; ‘Check C’ drops participants who picked “not applicable” in response to the party identification question. Income = “Differences in income in the United States are too large”; Redistribution = “It is the responsibility of the government to reduce the differences in income between people with high incomes and those with low incomes”; Vulnerable = “The most vulnerable in society are hit hardest by the coronavirus, also known as COVID-19”; Measures = “The measures taken against the coronavirus are more harmful than the virus itself.” *p < .05, **p < .01, ***p < .00 (two-tailed). To visualize our main findings, Fig. 3 plots predictive margins calculated from the regression models, comparing concerns about inequality and support for redistribution between participants in the treatment and control group by party identification. We find a significant difference between the controls and treated, indicative of a conditional treatment effect on participants’ belief that income inequality is too high, ranging from 0.28 points (95% CI, 0.03–0.51) among strong Democrats to 0.26 points for participants in the middle (95% CI, 0.07–0.53), and 0.54 points among Republicans (95% CI, 0.14–0.94). We also find positive differences between control and treatment group for Democrats (0.18 points) and strong Republicans (0.03 points), albeit not significantly different from zero at p < .05.Fig. 3 Beliefs about inequality by treatment and party identification. Note. Predictive margins based on OLS regression models including pretreatment attitudes and all control variables. Whiskers indicate the 95% confidence interval around each estimate. Fig. 3 We only find conditional treatment effects for participants’ support for government redistribution among Republicans, 0.72 points in size (95% CI, 0.21–1.22). Simply put, the informational treatment brings Republicans three-quarters up the way from “neither agree nor disagree” toward “somewhat agree.” These findings support the expectation that information provision impacts beliefs across partisan lines (Hypothesis 2B) and provide no support for the alternative expectation that the response to information is limited only to Democrats (Hypothesis 2A). Taken together, we find evidence of a shrinking ideological gap between moderate Republicans and Democrats following the provision of factual information about COVID-19's consequences for economic inequality. Among participants in the control group, the ideological gap in concerns about inequality is about 1 point between the partisan poles and 0.7 points when comparing moderate Republicans and Democrats. The partisan gap in support for redistribution is 0.9 and 1 point respectively. Our treatment reduces the gap between moderate Democrats and Republicans to 0.3 points on both topics (i.e. compared to 0.7 and 1); a gap that is not significantly different from zero at p < .05, reflective of an especially large attitudinal change among moderate Republicans. This finding however does not extend to those at the polar ends of the political spectrum. Comparing strong Democrats to strong Republicans in the treatment group, the polar ideological gap in concerns about inequality and support for redistribution increases to 1.2 and 1.5 points, respectively (compared to 1 and 0.9 in the control group). This finding adds an important qualification to our empirical support for Hypothesis 2B: provision of information affects beliefs on both sides of the political divide but may increase intraparty polarization. We return to this finding in our conclusion. Fig. 4 visualizes participants’ beliefs about COVID-19’s effects on the most vulnerable groups in society and the measures taken by government. For the former, we find a significant difference between the treated and controls, indicative of a conditional treatment effect of 0.43 points (95% CI, 0.07 - 0.85) among Republicans--almost half the way from “somewhat agree” to “agree”. For all other groups, we find positive differences between treated and controls, ranging from 0.10 points among participants in the middle, 0.23 for Democrats, 0.22 for strong Democrats, and 0.11 for strong Republicans, none of which however are significantly different from zero at p .05. Thus, we find a pattern of results comparable to those described above, lending no support for Hypothesis 2A and conditional support to Hypothesis 2B. Turning to participants’ beliefs about the negative consequences of government measures vis-à-vis the virus (Fig. 4 ), we find one statistically significant difference between the controls and treated: among strong Republicans, participants in the treatment group are more likely by 0.87 points (95% CI, 0.08–1.67) to believe that the measures taken by government are worse than the virus itself. Apparently, among these participants, 60 percent of which already believed the antidote to be worse than the illness (Fig. 1b), the informational treatment bolstered their concerns. This finding underlines the qualification we previously made with regard to the empirical support for Hypothesis 2B.Fig. 4 Beliefs about COVID-19 by treatment and party identification. Note. Predictive margins based on OLS regression models including pretreatment attitudes and all control variables. Whiskers indicate the 95% confidence interval around each estimate. Fig. 4 What does this mean for the ideological gap in beliefs about COVID-19-induced inequality and measures? Among control group participants, we find a partisan gap in the belief that COVID-19 has disproportionately affected the most vulnerable groups in society of 0.5 points both among moderates and strong partisans. The ideological gap in the belief that government measures have done more harm than the virus is 0.7 and 1.3 points, respectively. Our informational treatment reduces the gap in beliefs about COVID-19's consequences to 0.3 points between moderate Republicans and Democrats (not significantly different from zero at p < .05) but increases it to 0.6 points between the polar ends. Among moderates, concerning the role of government during the crisis we find a reduction in the attitudinal gap from 0.7 to 0.6 point (n.s.; p < .05). The gap between strong Democrats and strong Republicans however increases from 1.3 to 2.3, especially driven by the latter who become even more convinced that government COVID-19 measures have done more harm than the virus. We should note that this finding does not perfectly replicate in the robustness checks, which indicate similarly large coefficients for strong Republicans but report p-values just above the 5% level (Check A: 0.73, p = .07; Check B: 0.79, p = .06; Check C: 0.77, p = .06). 5 Conclusions Inspired by studies reporting surprisingly limited popular concern over steadily rising inequalities in recent decades (Breznau and Hommerich, 2019; Kenworthy and McCall, 2008; Luebker, 2014; Trump, 2017), this study asked whether a sudden and vast increase in inequalities in times of crisis (cf., Robinson et al., 2021) would make people's ideological beliefs more pliable than evidence from ‘normal’ times would suggest. Specifically, does the provision of factual information about COVID-19-induced inequalities make people more concerned and more supportive of income redistribution? We address this question by means of an original population-based survey experiment fielded in the least-likely case of the United States, renowned for its relatively high tolerance for income differences (Alesina et al., 2018) and partisan filtering of information (Bolsen et al., 2014). We find that Americans are split by partisan lines on each topic related to inequality. However, when exposed to the informational treatment, Americans across the ideological spectrum express 1) more concerns over economic inequality, 2) stronger support for income redistribution, and 3) stronger acknowledgement that the coronavirus has especially hurt the most vulnerable in society. These findings provide no support for the expectation that information about COVID-19-induced inequality only leads to belief change among Democrats. Instead, they lend support to the alternative Hypothesis that the provision of information in times of crisis may lead to belief change on both sides of the political divide. Research on the Great Recession in the United States (Margalit, 2013) and the Netherlands (Gidron and Mijs, 2019) documents growing support for redistribution especially among those who personally experienced economic hardship, reflecting self-interested concerns. Our findings suggest that a crisis like COVID-19 may fuel more broadly shared sociotropic concerns. Hence, further research into the specific conditions fueling both types of concern across the Atlantic would be most worthwhile. Turning to conditional treatment effects along ideological lines, in line with Bisgaard's study on the United Kingdom (2015) we find that even in the United States, moderate partisans are not far apart, assured they are provided with the same set of factual information. In the American case, the political divide is bridged by moderate Republicans who become markedly more supportive of redistribution following the informational treatment. However, while our informational treatment rendered non-significant the gap between moderates, it substantially increased the gap between moderate and strong Republicans. Moderate Republicans are receptive to new information and perspectives in times of crisis (cf. Swidler, 1986), whereas strong Republicans hold on to their political commitment or double down (cf. Nyhan and Reifler, 2010). We can think of three reasons for this intra-party polarization among Republicans. Perhaps our informational intervention, framed as non-partisan academic facts, rubbed strong Republicans the wrong way, given their relatively high levels of distrust of experts and the nonpartisan news media (Evans and Hargittai, 2020; Shepherd et al., 2020), in which this type of reporting is the default. Alternatively, strong Republicans could have attributed the steep rise in COVID-19-induced unemployment presented in the experimental condition as the consequence of government interventions, especially government shutdowns, hampering people from getting back to work. Such views typically go together both with their partisan position and demographic profile (as in other population-based surveys, strong Republicans in our sample were more likely than both moderate Republicans and all others to be male, white, Protestant, and have higher incomes and levels of education; see SI, Table S6 for details). Yet another factor may be that strong partisans -- Democrats and Republicans -- are simply more resistant to information that opposes their world view. Either way, we consider it less likely that information describing COVID-19-induced inequalities would invoke such polarization in other Western settings, as right-wing populist constituencies have been observed to be more pro-redistribution than those who prefer non-populist right-wing parties (De Koster et al., 2013). Given the ideological profile of the contemporary European left and right, information on cultural issues such as immigration and national sovereignty is more likely to incite polarization (Rydgren, 2008; Van Elsas et al., 2016). We welcome future research on the polarizing or unifying effects of informational treatments about inequalities in other settings. An important question that remains, concerns the scope and implications of our findings beyond the experimental setting. As the COVID-19 pandemic “does not appear to have fundamentally changed how subjects respond to treatments” in online experiments (Peyton et al., 2021), we are confident that exposure to information depicting the sudden and substantial increase in COVID-19-related inequality increases concerns among the bulk of the American population. The effect of information is likely to be greater in contexts with a lower tolerance of economic inequality. Yet, while our study exposes the public to the same set of information, in their daily lives, many Americans are exposed to partisan media mirroring their ideological profile and reinforcing their perspective (Bruine de Bruin et al., 2020). It follows that in contexts with a less polarized media landscape – such as the public broadcasting systems of various European countries (Mosca and Quaranta, 2016) – the public may be more uniformly informed about economic inequality, and consequently more likely to express concerns when confronted with information about its rampant rise. These suggest testable hypotheses for future research. Having established that most people – even the divided American public – express more concern about inequality when confronted with factual information, a subsequent question regards whether this finding translates to other contentious issues. For instance, can informational interventions produce a similar effect when targeting compliance with measures to mitigate the pandemic (see also Kelley and Evans, 2021)? Providing such information is challenging in a context where “[p]olitical leaders and media outlets on the right and left have sent divergent messages about the severity of the crisis, which could impact the extent to which Republicans and Democrats engage in social distancing and other efforts to reduce disease transmission” (Allcott et al., 2020, p. 1). Yet when information on measures to mitigate disease transmission is more uniformly available across the political landscape, there is good reason to believe the partisan gap can be bridged. To this point, Druckman et al. (2021, p. 36) observed that closing the partisan gap on the use of masks followed “changing rhetoric by Republican elites—including President Trump—to follow the Democratic perspective on mask wearing.” All in all, whereas steadily rising inequalities have sparked remarkably little public concern in recent decades, uniformly confronting the public with factual information describing the economic consequences of COVID-19 makes even the polarized American public more worried about inequality and more supportive of income redistribution. The strikingly uniform pattern of responses among moderates on both sides of the political divide suggest that disagreement over inequality may be rooted not in fundamentally incompatible worldviews but in different perceptions of how things are, which prove pliable through the provision of information. At the same time, our findings suggest that a crisis like COVID-19 may upend ideological rifts among moderates, while increasing their salience on the right of the political spectrum. We expect information describing a sudden and substantial increase in inequalities to be an even more likely source of concern in many European countries, where it is less likely to inspire attitudinal polarization at the political fringe. Funding Jonathan Mijs received funding from a 10.13039/100010665 Marie Skłodowska-Curie Individual Fellowship, EU Commission Horizon 2020 Grant no. 88296 and a Veni grant from the 10.13039/501100003246 Nederlandse Organisatie voor Wetenschappelijk Onderzoek (NWO), grant no. VI.Veni.201S.003. Willem de Koster and Jeroen van der Waal received financial support through Vidi grants from the 10.13039/501100003246 NWO , nos. 016.Vidi.185.207 and 452-17-009. Data availability Replication data and Stata scripts have been made available through the Open Science Framework to allow for independent verification of our findings: https://osf.io/ub538. Appendix A Supplementary data The following are the Supplementary data to this article:Multimedia component 1 Multimedia component 1 Multimedia component 2 Multimedia component 2 Appendix A Supplementary data to this article can be found online at https://doi.org/10.1016/j.ssresearch.2021.102692. ==== Refs References Alesina A. Stantcheva S. Teso E. Intergenerational mobility and preferences for redistribution Am. Econ. Rev. 108 2018 521 554 10.1257/aer.20162015 Allcott H. Boxell L. Conway J. Gentzkow M. Thaler M. Yang D. Polarization and public health: partisan differences in social distancing during the coronavirus pandemic J. Publ. Econ. 191 2020 104254 Aronow P.M. Baron J. Pinson L. A note on dropping experimental subjects who fail a manipulation check Polit. Anal. 27 2019 572 589 Athreya K. Mather R. Mustre-del-Rio J. Sánchez J.M. Employment Vulnerability and (Financial) Pre-existing Conditions 2020 Federal Reserve Bank of Richmond Atkinson A.B. Piketty T. Saez E. Top incomes in the long run of history J. Econ. Lit. 49 2011 3 71 10.1257/jel.49.1.3 Beall A.T. Hofer M.K. Schaller M. Infections and elections: did an Ebola outbreak influence the 2014 U.S. Federal elections (and if so, how)? Psychol. Sci. 27 2016 595 605 10.1177/0956797616628861 26976083 Berrebi C. Klor E.F. Are voters sensitive to terrorism? Direct evidence from the Israeli electorate Am. Polit. Sci. Rev. 102 2008 279 301 10.2307/27644521 Bird R. Ritter Z. Is the Media Creating Division on COVID-19 Health Practices? 2020 Gallup Washington, DC Bisgaard M. Bias will find a way: economic perceptions, attributions of blame, and partisan-motivated reasoning during crisis J. Polit. 77 2015 849 860 10.1086/681591 Bolsen T. Druckman J.N. Cook F.L. The influence of partisan motivated reasoning on public opinion Polit. Behav. 36 2014 235 262 10.1007/s11109-013-9238-0 Bradley D. Huber E. Moller S. Nielsen F. Stephens J.D. Distribution and redistribution in postindustrial democracies World Polit. 55 2003 193 228 Breznau N. Hommerich C. No generalizable effect of income inequality on public support for governmental redistribution among rich democracies 1987–2010 Soc. Sci. Res. 81 2019 170 191 10.1016/j.ssresearch.2019.03.013 31130195 Bruine de Bruin W. Saw H.-W. Goldman D.P. Political polarization in US residents' COVID-19 risk perceptions, policy preferences, and protective behaviors J. Risk Uncertain. 61 2020 177 194 10.1007/s11166-020-09336-3 33223612 Bureau of Labor Statistics Labor Force Statistics from the Current Population Survey 2020 U.S. Bureau of Labor Statistics Washington, DC Cappelen A.W. Falch R. Sorensen E.O. Tungodden B. Wezerek G. Opinion | what Do You Owe Your Neighbor? the Pandemic Might Change Your Answer 2020 N. Y. Times CBO The Distribution of Household Income, 2016 | Congressional Budget Office 2019 Congressional Budget Office Washington, DC Civiqs Coronavirus: Outbreak Concern 2020 Civiqs Oakland, CA Clifford S. Sheagley G. Piston S. Increasing precision without altering treatment effects: repeated measures designs in survey experiments Am. Polit. Sci. Rev. 2021 10.1017/S0003055421000241 (in press) Collins C. U.S. Billionaires got 62 percent richer during pandemic They’re Now up $1.8 Trillion 2021 Institute for Policy Studies Conway L.G. III Woodard S.R. Zubrod A. Chan L. Why are conservatives less concerned about the coronavirus (COVID-19) than liberals? Comparing political, experiential, and partisan messaging explanations Pers. Indiv. Differ. 183 2021 111 124 Cox J. The Upcoming Job Losses Will Be unlike Anything the US Has Ever Seen 2020 CNBC Cox R.H. The social construction of an imperative: why welfare reform happened in Denmark and The Netherlands but not in Germany World Polit. 53 2001 463 498 17595731 Dalton R.J. The quantity and the quality of party systems: party system polarization, its measurement, and its consequences Comp. Polit. Stud. 41 2008 899 920 10.1177/0010414008315860 De Koster W. Achterberg P. Van der Waal J. The new right and the welfare state: the electoral relevance of welfare chauvinism and welfare populism in The Netherlands Int. Polit. Sci. Rev. 34 2013 3 20 DeRensis H. The coronavirus economic crisis: the next Great depression? Natl 2020 Interest March 20 Druckman J.N. Klar S. Krupnikov Y. Levendusky M. Ryan J.B. Affective polarization, local contexts and public opinion in America Nat. Hum. Behav. 5 2021 28 38 10.1038/s41562-020-01012-5 33230283 Eadeh F.R. Chang K.K. Can threat increase support for liberalism? New insights into the relationship between threat and political attitudes Soc. Psychol. Personal. Sci. 11 2020 88 96 10.1177/1948550618815919 Evans J.H. Hargittai E. Who doesn't trust fauci? The public's belief in the expertise and shared values of scientists in the COVID-19 pandemic Socius 6 2020 10.1177/2378023120947337 2378023120947337 Flaxman S. Goel S. Rao J.M. Filter bubbles, echo chambers, and online news consumption Publ. Opin. Q. 80 2016 298 320 10.1093/poq/nfw006 Gerring J. Is there a (viable) crucial-case method? Comp. Polit. Stud. 40 2007 231 253 Gidron N. Mijs J.J.B. Do changes in material circumstances drive support for populist radical parties? Panel data evidence from The Netherlands during the Great recession, 2007–2015 Eur. Socio Rev. 2019 10.1093/esr/jcz023 Gollwitzer A. Martel C. Brady W.J. Pärnamets P. Freedman I.G. Knowles E.D. Van Bavel J.J. Partisan differences in physical distancing are linked to health outcomes during the COVID-19 pandemic Nat. Hum. Behav. 4 2020 1186 1197 33139897 Gordon R.J. The Rise and Fall of American Growth 2016 Princeton University Press Princeton, NJ Grasso M. Klicperová-Baker M. Koos S. Kosyakova Y. Petrillo A. Vlase I. The impact of the coronavirus crisis on European societies. What have we learnt and where do we go from here? – introduction to the COVID volume Eur. Soc. 2021 1 31 10.1080/14616696.2020.1869283 (in press) Grossman G. Kim S. Rexer J.M. Thirumurthy H. Political partisanship influences behavioral responses to governors' recommendations for COVID-19 prevention in the United States Proc. Natl. Acad. Sci. Unit. States Am. 117 2020 24144 24153 10.1073/pnas.2007835117 Havey N.F. Partisan public health: how does political ideology influence support for COVID-19 related misinformation? J. Comput. Soc. Sci. 3 2020 319 342 33163686 ISSP Research Group International Social Survey Programme 2019 Social Inequality V. Final Questionnaire 2018 GESIS Cologne Jerit J. Zhao Y. Political misinformation Annu. Rev. Polit. Sci. 23 2020 77 94 10.1146/annurev-polisci-050718-032814 Jost J.T. Glaser J. Kruglanski A.W. Sulloway F.J. Political conservatism as motivated social cognition Psychol. Bull. 129 2003 339 375 10.1037/0033-2909.129.3.339 12784934 Kelley J. Evans M.D.R. Legitimate earnings inequality and national welfare commitment: correspondence between economic institutions and the pay 80,000+ people in 30 nations think legitimate for ordinary jobs and for elite jobs Soc. Sci. Res. 94 2021 102446 10.1016/j.ssresearch.2020.102446 33648680 Kelly N.J. Enns P.K. Inequality and the dynamics of public opinion: the self-reinforcing link between economic inequality and mass preferences Am. J. Polit. Sci. 54 2010 855 870 10.1111/j.1540-5907.2010.00472.x Kenworthy L. McCall L. Inequality, public opinion and redistribution Soc. Econ. Rev. 6 2008 35 68 10.1093/ser/mwm006 Kenworthy L. Pontusson J. Rising inequality and the politics of redistribution in affluent countries Perspect. Polit. 3 2005 449 471 Kozlowski A.C. Murphy J.P. Issue alignment and partisanship in the American public: revisiting the ‘partisans without constraint’ thesis Soc. Sci. Res. 94 2021 102498 10.1016/j.ssresearch.2020.102498 33648681 Kreps S.E. Kriner D.L. Model uncertainty, political contestation, and public trust in science: evidence from the COVID-19 pandemic Sci. Adv. 6 2020 eabd4563 10.1126/sciadv.abd4563 32978142 Kuklinski J.H. Quirk P.J. Jerit J. Schwieder D. Rich R.F. Misinformation and the currency of democratic citizenship J. Polit. 62 2000 790 816 10.1111/0022-3816.00033 Kuziemko I. Norton M.I. Saez E. Stantcheva S. How elastic are preferences for redistribution? Evidence from randomized survey experiments Am. Econ. Rev. 105 2015 1478 1508 10.1257/aer.20130360 Lee K.S. Fujita Y. Economic recession and the nature and pace of social change in Japan Soc. Sci. Res. 40 2011 784 795 10.1016/j.ssresearch.2011.01.002 Lin W. Agnostic notes on regression adjustments to experimental data: reexamining Freedman's critique Ann. Appl. Stat. 7 2013 295 318 10.1214/12-AOAS583 Loveless M. Whitefield S. Being unequal and seeing inequality: explaining the political significance of social inequality in new market democracies Eur. J. Polit. Res. 50 2011 239 266 10.1111/j.1475-6765.2010.01929.x Luebker M. Income inequality, redistribution, and poverty: contrasting rational choice and behavioral perspectives Rev. Income Wealth 60 2014 133 154 10.1111/roiw.12100 Margalit Y. Explaining social policy preferences: evidence from the Great recession Am. Polit. Sci. Rev. 107 2013 80 103 10.1017/S0003055412000603 McCall L. Burk D. Laperrière M. Richeson J.A. Exposure to rising inequality shapes Americans' opportunity beliefs and policy support Proc. Natl. Acad. Sci. Unit. States Am. 114 2017 9593 9598 10.1073/pnas.1706253114 Mijs J.J.B. Visualizing Belief in Meritocracy, 1930-2010 2018 Socius 4 Mijs J.J.B. The paradox of inequality: income inequality and belief in meritocracy go hand in hand Socio-Economic Review 19 1 2021 7 35 10.1093/ser/mwy051 Mora G.C. Schickler E. Paschel T. Perceptions of Inequality and the Pandemic Vary Drastically Among Californians 2020 Inst. Gov. Stud. Release 2020-08 Mosca L. Quaranta M. News diets, social media use and non-institutional participation in three communication ecologies: comparing Germany, Italy and the UK Inf. Commun. Soc. 19 2016 325 345 Mummolo J. News from the other side: how topic relevance limits the prevalence of partisan selective exposure J. Polit. 78 2016 763 773 10.1086/685584 Naumann E. Buss C. Bähr J. How unemployment experience affects support for the welfare state: a real panel approach Eur. Socio Rev. 32 2016 81 92 Neimanns E. Busemeyer M.R. Garritzmann J.L. How popular are social investment policies really? Evidence from a survey experiment in eight western European countries Eur. Socio Rev. 34 2018 238 253 10.1093/esr/jcy008 Nyhan B. Reifler J. When corrections fail: the persistence of political misperceptions Polit. Behav. 32 2010 303 330 10.1007/s11109-010-9112-2 Oude Groeniger J. Noordzij K. van der Waal J. de Koster W. Dutch COVID-19 lockdown measures increased trust in government and trust in science: a difference-in-differences analysis Soc. Sci. Med. 275 2021 113819 10.1016/j.socscimed.2021.113819 33725488 Packer G. America's plastic hour is upon us Atlantic 2020 48 57 Palan S. Schitter C. Prolific.ac—a subject pool for online experiments J. Behav. Exp. Finance 17 2018 22 27 10.1016/j.jbef.2017.12.004 Pechar E. Bernauer T. Mayer F. Beyond political ideology: the impact of attitudes towards government and corporations on trust in science Sci. Commun. 40 2018 291 313 10.1177/1075547018763970 Peyton K. Huber G.A. Coppock A. The generalizability of online experiments conducted during the COVID-19 pandemic J. Exp. Polit. Sci. 2021 10.1017/XPS.2021.17 Robinson L. Schulz J. Ball C. Chiaraluce C. Dodel M. Francis J. Huang K.-T. Johnston E. Khilnani A. Kleinmann O. Cascading crises: society in the age of COVID-19 Am. Behav. Sci. 2021 00027642211003156 Rydgren J. Immigration sceptics, xenophobes or racists? Radical right-wing voting in six West European countries Eur. J. Polit. Res. 47 2008 737 765 Sachweh P. Sthamer E. Why do the affluent find inequality increasingly unjust? Changing inequality and justice perceptions in Germany, 1994–2014 Eur. Socio Rev. 35 2019 651 668 10.1093/esr/jcz024 Schüller S. The 9/11 conservative shift Econ. Lett. 135 2015 80 84 10.1016/j.econlet.2015.07.031 Schwartz N.D. ‘Nowhere to Hide’ as Unemployment Permeates the Economy 2020 N. Y. Times Shepherd H. MacKendrick N. Mora G.C. Pandemic politics: political worldviews and COVID-19 beliefs and practices in an unsettled time Socius 6 2020 10.1177/2378023120972575 2378023120972575 Suhay E. Klasnja M. Rivero G. Ideology of affluence: rich Americans' explanations for inequality and attitudes toward redistribution J. Polit. 2020 10.1086/709672 Swidler A. Culture in action: symbols and strategies Am. Socio. Rev. 51 1986 273 286 10.2307/2095521 Tierney K.J. From the margins to the mainstream? Disaster research at the crossroads Annu. Rev. Sociol. 33 2007 503 525 10.1146/annurev.soc.33.040406.131743 Trump K.-S. Income inequality influences perceptions of legitimate income differences Br. J. Polit. Sci. 48 2017 929 952 10.1017/S0007123416000326 Van der Waal J. Achterberg P. Houtman D. Class is not dead—it has been buried alive: class voting and cultural voting in postwar western societies (1956–1990) Polit. Soc. 35 2007 403 426 Van Elsas E.J. Hakhverdian A. Van der Brug W. United against a common foe? The nature and origins of Euroscepticism among left-wing and right-wing citizens W. Eur. Polit. 39 2016 1181 1204 Wagner‐Pacifici R. Theorizing the restlessness of events Am. J. Sociol. 115 2010 1351 1386 10.1086/651299 Waterfield S. A list of essential workers that we should thank and support during the coronavirus pandemic Newsweek 2020

35400387

PMC9754060

NO-CC CODE

2022-12-16 23:26:25

no

Soc Sci Res. 2022 May 21; 104:102692

utf-8

Soc Sci Res

10.1016/j.ssresearch.2021.102692

oa_other

==== Front Soc Sci Res Soc Sci Res Social Science Research 0049-089X 1096-0317 The Authors. Published by Elsevier Inc. S0049-089X(21)00169-1 10.1016/j.ssresearch.2021.102692 102692 Article Belief change in times of crisis: Providing facts about COVID-19-induced inequalities closes the partisan divide but fuels intra-partisan polarization about inequality Mijs Jonathan J.B. ab∗ de Koster Willem b van der Waal Jeroen b a Department of Sociology, Boston University, 100 Cummington Mall, Boston, MA, 02215, United States b Department of Public Administration and Sociology, Erasmus University Rotterdam, PO Box 1738, DR Rotterdam, 3000, the Netherlands ∗ Corresponding author 21 12 2021 5 2022 21 12 2021 104 102692102692 11 5 2021 4 11 2021 18 12 2021 © 2021 The Authors 2021 Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. Population-based survey research demonstrates that growing economic divides in Western countries have not gone together with increased popular concern about inequality. Extant explanations focus on ‘misperception’: people generally underestimate the extent of inequality and overestimate society's meritocratic nature. However, scholarly attempts to correct people's misperceptions have produced mixed results. We ask whether COVID-19, by upending everyday life, has made people responsive to information about inequality, even if that entails crossing ideological divides. We field an original survey experiment in the United States, a least-likely case of belief change, given high levels of inequality and partisan polarization. Our informational treatment increases (1) concerns over economic inequality, (2) support for redistribution, and (3) acknowledgement that COVID-19 has especially hurt the most vulnerable. Information provision renders non-significant the partisan gap between moderate Democrats and Republicans but increases that between moderate and strong Republicans. We discuss our findings' implications and suggestions for further research. Keywords COVID-19 Inequality Information provision Political polarization Support for redistribution ==== Body pmc1 Introduction The steep rise in economic inequality starting in the 1970s presents a puzzle to scholars of public opinion: while some studies describe growing support for government intervention (Kenworthy and Pontusson, 2005), in many countries we see surprisingly little attitudinal evidence of concern or calls for income redistribution among the public at large (Luebker, 2014; Trump, 2017; Bradley et al., 2003; Loveless and Whitefield, 2011; Breznau and Hommerich, 2019). Similarly, while some research documents a growing sense of injustice among certain segments of society (for the case of Germany, see: Sachweh and Sthamer, 2019), large-scale survey research reports stable or strengthening popular belief in meritocracy (Kelly and Enns, 2010; Kenworthy and McCall, 2008; Mijs, 2018, Mijs, 2021; Suhay et al., 2020). These patterns are typically explained by widespread ‘misperception’ of the nature and extent of economic disparities: many people confidently hold incorrect convictions and perceptions, which keeps them from seeing the full extent and non-meritocratic nature of economic inequality (Jerit and Zhao, 2020; Kuklinski et al., 2000; McCall et al., 2017). Whereas misperceptions are endemic, addressing them through the provision of factual information is complicated by ideological camps' distrust in science, experts and government (Pechar et al., 2018), different exposure to news and information (Flaxman et al., 2016; Mummolo, 2016), and, crucially, by partisan motivated reasoning in a polarized political landscape: “individuals interpret information through the lens of their party commitment” (Bolsen et al., 2014, p. 235). Evidence for attitudinal effects of informational interventions is mixed. While learning about high levels of inequality generally raises concerns about inequality (McCall et al., 2017), it typically generates support for redistribution only among people already positively inclined and trusting in government (Alesina et al., 2018; Kuziemko et al., 2015). Concerns over economic disparities might hardly be affected by COVID-19-induced surges in inequality, as the pandemic continues to split the public along partisan lines (Druckman et al., 2021; Grasso et al., 2021; Kreps and Kriner, 2020; Shepherd et al., 2020). However, research on past crises suggests that such “unsettled times” (Swidler, 1986) may inspire belief change (Bisgaard, 2015; Gidron and Mijs, 2019; Lee and Fujita, 2011; Margalit, 2013; Naumann et al., 2016). This paper investigates belief change in times of the COVID-19 crisis. Specifically, we ask whether information on COVID-19-induced inequalities heightens the public's concern about inequity and strengthens support for government intervention, and how the uptake of information is shaped by partisanship. Following fruitful prior empirical applications (e.g. Neimanns et al., 2018), we leverage a survey experiment methodology to analyze the effects, if any, of factual information describing the impact of the pandemic on people's beliefs about inequality, COVID-19, and the role of government. We field our study in the United States, which we consider a strategic least likely case (Gerring, 2007) of belief change. The US stands out from other advanced economies because of its high levels of economic inequality, coupled with relatively limited public support for income redistribution and deep political polarization, as demonstrated by extant research utilizing population-based surveys (Alesina et al., 2018; Atkinson et al., 2011; Kozlowski and Murphy, 2021). Hence, by fielding our study in the US, we aim to make a conservative assessment of the potential of informational interventions. Our results suggest that learning about COVID-19-induced inequalities inspires (1) stronger acknowledgement that COVID-19 has especially hurt the most economically vulnerable, (2) more concerns over economic inequality, and, crucially, (3) stronger support for income redistribution. Importantly, conditional treatment effects by partisanship reveal larger effects among moderate Republicans. Consequently, our informational treatment renders non-significant the partisan gap between moderate Democrats and Republicans, while the gap between moderate and strong Republicans widens. Thus, our findings suggest that uniform information on a crisis can increase concerns about inequality among substantial parts of the population. It can even overcome attitudinal gaps between moderate partisans in a country renowned for widespread polarization and relatively high tolerance of inequality. In what follows we situate our study in the broader literature on crises and political polarization before discussing our research design and results. We conclude by discussing how our findings speak to other country contexts and provide suggestions for further research. 2 Belief change in times of crisis Past research documents how tumultuous times can cause a shift in a conservative (Schüller, 2015) or liberal direction (Eadeh and Chang, 2020). For instance, scholars have documented how domestic and international crises, from the Ebola outbreak in the US (Beall et al., 2016) to terrorist attacks on American soil (Schüller, 2015) and abroad (Berrebi and Klor, 2008), have shifted public opinion in a conservative direction. Such a conservative response to systemic threat arguably reflects a heightened need to protect the status quo—no matter its unequal nature (Jost et al., 2003). Recent surveys suggest that a similar pattern may be unfolding today: Cappelen et al. (2020), in an unpublished survey of 8,000 Americans reported in The New York Times, find that people who were primed to think about the impact of the virus on their community cared more about their country and cared less about inequality. Other research suggests that the very scope of a crisis like the COVID-19 pandemic may make it a ‘common enemy,’ which may instead increase support for government intervention rather than retrenchment (cf. Oude Groeniger et al., 2021). Eadeh and Chang (2020) suggest that a crisis can cause a 'liberal turn' if the threat concerns policy areas like health care and the environment, where liberal politicians are perceived to be more competent. The New Deal, for instance, can be seen as a direct response to the Great Depression (Gordon, 2016) and, more recently, the Great Recession led the Obama administration to the greatest redistributive effort in three decades (CBO, 2019). In this particular historical moment, public appreciation of ‘essential workers’ may go together with changing perceptions, attitudes and policies addressing inequality (Waterfield, 2020). President Biden's Build Back Better Framework, arguably, is a step in that direction. At the same time, it is unlikely that Americans across the political divide have experienced and understood the pandemic in the same way. Such is the finding of a Californian survey on the perceived impact of the COVID-19 pandemic on economic inequality: over 80 percent of Democrats but just 40 percent of Republicans believed economic inequality had increased (Mora et al., 2020; Shepherd et al., 2020). A similar partisan divide characterizes Americans’ concerns about the coronavirus more generally—and the gap is widening, as suggested by weekly surveys between early and late 2020 (Civiqs, 2020).Hypothesis 1 Americans’ beliefs about COVID-19, economic inequality and the role of government are polarized along party lines. Compounding the ideological divide, the public's political response to COVID-19 likely depends on how the crisis is understood by different ideological camps (Bird and Ritter, 2020; Cox, 2001; Tierney, 2007). Theorizing on partisan motivated reasoning (Bolsen et al., 2014) suggests that information about COVID-19-induced inequalities is filtered through a partisan lens, making some parts of the population more receptive to it than others. Republicans are politically motivated to accept economic inequalities as deserved, to oppose government intervention in the economy, and to be less supportive and compliant with COVID-19-related government interventions more specifically (cf. Conway et al., 2021; Gollwitzer et al., 2020). Conversely, Democrats traditionally express more concerns about inequality, and are more supportive of income redistribution (Kozlowski and Murphy, 2021; Pechar et al., 2018) and of public spending on COVID-19-related healthcare (Gollwitzer et al., 2020). It follows that provision of the same factual information may have a different impact across partisan lines, leaving Republicans unaffected while resonating with Democrats who may already be so inclined (Grossman et al., 2020).Hypothesis 2A Information about COVID-19-induced inequalities heightens concerns about inequality and strengthens support for government intervention only among Democrats. An alternative perspective considers a crisis, more fundamentally, as a “plastic hour” (Gershom, cited in Packer, 2020, p. 50) in which taken-for-granted practices, policies, and attitudes are upended. Could such 'unsettled times' (Swidler, 1986) constitute an 'event' that can shock or rupture political divides (Wagner-Pacifici, 2010)? Specifically, may times like these make people responsive to information about inequality, even if doing so means crossing ideological rifts? Research from the United Kingdom suggests the 2008 Great Recession did just that. Bisgaard (2015, p. 840) describes a pre-crisis partisan gap in perceptions about the economy which “evaporates” during the crisis, as even the staunchest partisans share the dire diagnosis of their country's economic state, even if they disagree about where to lay blame. Research on the Netherlands (Gidron and Mijs, 2019), Germany (Naumann et al., 2016), and the United States (Margalit, 2013) similarly describes growing support for redistribution across the political spectrum in times of crisis. Given its deep impact across society as a combined economic and public health crisis, the COVID-19 pandemic arguably constitutes a greater ‘rupture’ in everyday life than most other crises: scholars observe a “tsunami of change” and note that “the unusual conditions of the pandemic – unlike other crises – have impacted almost every facet of our lives” (Robinson et al., 2021, pp. 1608–9). As such, it may produce a particular ‘plastic’ moment.Hypothesis 2B Information about COVID-19-induced inequalities heightens concerns about inequality and strengthens support for government intervention on both sides of the political divide. 3 Data and Methods 3.1 Survey design Previous studies suggest that the most effective informational treatments are designed as non-partisan, cognitively light (Alesina et al., 2018), informational interventions designed to ‘shock’ participants' belief system (Kuziemko et al., 2015). Incorporating these insights, we developed an ‘omnibus treatment’ (cf. Kuziemko et al., 2015) describing COVID-19's economic consequences (Supplementary Information, Fig. S1). Participants were shown a graph of the number of Americans filing for unemployment between January 2020 and July 2020. The graph is accompanied by facts taken from various trusted, nonpartisan, sources, which (1) highlight the total number of people who filed for unemployment (cf. Bureau of Labor Statistics, 2020), (2) introduce the prognosis that this crisis will have a larger economic effect than any other crisis in recent history (cf. Cox, 2020; DeRensis, 2020; Schwartz, 2020), (3) emphasize its disproportionate effects on low and middle-income workers (cf. Athreya et al., 2020), and (4) inform participants that, meanwhile, some of the wealthiest Americans' fortunes have significantly grown (cf. Collins, 2021). Participants in the control condition were presented an unrelated but similarly looking graph depicting what share of different age groups are getting enough exercise, accompanied with facts about the positive health effects of physical exercise and stating the share of youth and adults that meets the recommended level of sports and exercise (SI, Fig. S2). The treatment was embedded in a between-subject survey design incorporating pretreatment and post-treatment questions. As in a standard between-subject design, we identify the treatment effect as the difference in post-treatment responses between participants in the treatment and control condition. Incorporating pretreatment questions that are distinct from but correlated with our post-treatment questions produces higher precision and more statistical power than a standard between-subjects design (Clifford et al., 2021; Lin, 2013). Specifically, we asked three questions which are correlated with the post-treatment questions about inequality, COVID-19 and the role of government (0.21≤ r ≤ 0.60) and include these as pretreatment controls in regression models estimating the treatment effect. This means that participants in both the control and treatment condition are introduced to the topic of inequality prior to our measurement of their post-treatment beliefs. As such, our design produces a conservative estimate of the effect of information over and above a baseline level of inequality priming. 3.2 Data We set to recruit 1,000 participants using a quota sample provided by Prolific Academic stratified by sex, age and race/ethnicity to match US Census Current Population Statistics. We recruited participants between August 5 and August 11, 2020, through Prolific Academic. Prolific is a survey firm specializing in social science research, founded by academics in Oxford, UK. It has worked with researchers at top institutions around the world and compares favorably to other survey firms that offer high-quality alternatives to Amazon Mechanical Turk (Palan and Schitter, 2018). We fielded our survey experiment with Prolific's active panel of 138,363 participants based in the US. Panelists are registered after verification of a valid e-mail address, phone number and payment method. Each panelist is assigned a unique identifier, matched with self-reported basic demographic information. They receive compensation for each survey completed, after an evaluation of their survey responses. Panelists flagged for low-quality responses more than once are removed from the panel. We obtained a sample of 1,003 participants. Ten participants (one percent) did not complete the survey. New participants were recruited in their place. The final sample matches population statistics on race and gender but skews slightly toward a younger demographic (SI, Table S3). Participants were randomly assigned to either the control (n = 500) or treatment condition (n = 503). Based on power calculations, we ensured that treatment and control group had 500 participants per condition to get a power of 0.9 when the Cohen's d = 0.2. We obtain almost perfect post-allocation balance between participants in the control and treatment group on key dimensions (SI, Table S4). We took several measures to secure the quality of our research. First, to accommodate people differently affected by COVID-19, working and not working, with and without caring duties, we provided an extended window, spanning two working days and a weekend day, during which participants could take the survey. Second, we designed the survey to be short: the median time of completion was 11 min. Third, we tested our questions and treatment design in two pilot surveys (n = 100 and n = 150). Fourth, to minimize selection bias, we gave our survey a non-descript name (“Social topics in the United States”) and set compensation at a relatively generous $2.50, corresponding to an hourly rate of approximately $14. Fifth, we include a post-treatment attention check, by asking participants which informational treatment they were given (coronavirus, exercise, dining, don't know). Only eighteen participants (1.8 percent) failed the check, which indicates that respondents were generally attentive. Those who failed the check were kept in the analysis so as not to induce bias (Aronow et al., 2019). The main results are qualitatively equivalent to including an attention check dummy (Table 2; Check B). Finally, we ran checks for survey straightlining, but found no concerning patterns in our data. We have made our data and code publicly available on the Open Science Foundation platform. 3.3 Measures We focus on four attitudinal variables about inequality, COVID-19, and the role of government, measured on a 7-point scale ranging from “Strongly disagree” to “Strongly agree”: 1) “differences in income in the United States are too large,” 2) “it is the responsibility of the government to reduce the differences in income between people with high incomes and those with low incomes,” 3) “the most vulnerable in society are hit hardest by the coronavirus, also known as COVID-19,” and 4) “the measures taken against the coronavirus are more harmful than the virus itself.” Questions 1 and 2 are adopted from the International Social Survey Programme Social Inequality module to allow for a direct comparison with international research (ISSP Research Group, 2018). Question 3 and 4 are original questions designed to capture strong sentiments about the impact of the pandemic as expressed on different sides of the political aisle: whereas the former taps into to concerns about COVID-19-related inequality most frequently found among Democrats (Mora et al., 2020; Shepherd et al., 2020), the latter reflects a longstanding concern of economic conservatives regarding the unemployment-inducing nature of government interference in the economy, more typically found among Republicans (Bruine de Bruin et al., 2020; Van der Waal et al., 2007). Because of indications of substantial intra-party variation in COVID-19-related attitudes (e.g., Havey, 2020), we use a more fine-grained measure of partisanship than most COVID-19 studies to date. To concisely discern weak from strong partisans, we asked for participants’ self-placement on a 10-point scale ranging from strong Democrat to strong Republican, using the middle as a default starting position (cf. Dalton, 2008). In the analyses below, we compare participants strongly identifying as Democrat (0–1) to those identifying as Democrat (2–3), those identifying as Republican (7–8), strongly identifying as Republican (9–10) and those in the middle (4–6). Fifty participants (5 percent) opted out of the question and were grouped with the middle category. Excluding these participants, in a robustness check, does not change our findings (Table 2; Check C). As a final robustness check, we replicate our main analyses by including additional pretreatment controls for age, gender, education, parental education, marital status, household income, employment status, self-placement on the social ladder, financial assets, religion and the date the survey was taken (for descriptive statistics, see SI, Table S5). Doing so, we obtained qualitatively equivalent results (Table 2; Check A). Table 1 provides descriptives for key dependent and independent variables.Table 1 Sample descriptives (n = 1,003). Table 1Variable Mean SD Dependent variables Differences in income too large (1–7) 5.74 1.37 Government responsibility to reduce differences in income (1–7) 4.98 1.80 Most vulnerable are hit hardest by COVID-19 (1–7) 6.11 1.14 Measures taken against COVID-19 are more harmful than the virus (1–7) 2.82 1.91 Independent variables Treatment assignment (0/1) 0.50 Party identification  Strong Democrat 0.30  Democrat 0.19  Neither 0.33  Republican 0.11  Strong republican 0.08 Pretreatment controls Society is fair when hard-working people earn more (1–7) 4.86 1.45 Racial diversity makes America stronger (1–7) 5.75 1.48 For society to be fair, income differences should be small (1–7) 4.87 1.61 4 Results 4.1 Ideological divide in beliefs about inequality, COVID-19 and role of government To address Hypothesis 1 regarding the political polarization of beliefs about inequality, COVID-19 and the role of government, we first present histograms to visualize the distribution of responses by party identification. Fig. 1 a describes a monotonic partisan divide, both in concerns about income inequality and support for redistribution, which corroborates Hypothesis 1. Participants identifying as strong Democrats are consistently more concerned about inequality than participants in the middle, and, in turn, than Republicans and strong Republicans (6.4 > 6.1 > 5.5 > 5.0 > 4.4; significantly different at p < .05), and more supportive of income redistribution (6.0 > 5.5 > 4.4 > 4.0 ≥ 3.6; all but the latter two are significantly different at p < .05).Fig. 1 Histogram of beliefs about (a) inequality and (b) COVID-19 by party identification. Note. Bars indicate the percentage of responses across response categories within each group. Question wording: “Differences in income in the United States are too large”; “It is the responsibility of the government to reduce the differences in income between people with high incomes and those with low incomes”; “The most vulnerable in society are hit hardest by the coronavirus (also known as COVID-19)”; and “The measures taken against the coronavirus are more harmful than the virus itself.” Fig. 1 Fig. 1b shows that strong Democrats and Democrats are also monotonically more convinced than Republicans and participants in the middle that the coronavirus has disproportionally affected vulnerable populations (6.5 > 6.3 > 5.9 > 5.5 ≥ 5.5; all contrasts but the latter two are significantly different at p < .05). Conversely, (strong) Republicans are more likely than (strong) Democrats to think that the measures taken to combat the virus have been more harmful than the virus itself, but Republicans not significantly more so than participants in the middle (2.0 < 2.4 < 3.2 ≤ 3.5 < 4.6). Taken together, our results describe a partisan divide both in perceptions of inequality and the coronavirus and in attitudes about the role the government, in line with Hypothesis 1. Below, we first consider the general effect of provision of information, before testing our Hypotheses regarding how partisanship shapes the effect of information. 4.2 Treatment effect of information describing COVID-19-induced inequalities To evaluate the effect of the informational treatment on participants’ beliefs about inequality, COVID-19 and the role of government, we identify the average treatment effect of information by estimating OLS regressions, modeling each outcome as a function of the information treatment, and the pretreatment controls described above (Table 2). Fig. 2 plots average marginal effects calculated from the regression models and shows a positive treatment effect for three out of four attitudes. The informational treatment is associated with a 0.26-point (95% CI, 0.12–0.40) increase in participants' belief that income inequality is too high, a 0.19-point (95% CI, 0.01–0.36) increase in support for government redistribution, and a 0.20-point (95% CI, 0.07–0.33) increase in the belief that COVID-19 has disproportionately affected society's most vulnerable groups. We do not find a significant treatment effect (p < .05) for participants' belief that government measures are worse than the virus.Fig. 2 Treatment effect on beliefs about inequality, COVID-19, and the role of government. Note. Average marginal effects are estimated from OLS regression models including pretreatment attitudes and all control variables. Whiskers indicate the 95% confidence interval around each estimate. Fig. 2 All in all, even in a strongly polarized country known for its high level of economic inequality and comparatively low levels of public concern about this, we find a substantively meaningful effect of information provision across three dimensions of inequality, between a fifth and a quarter of a point on a 7-point attitudinal scale. But how does factual information about COVID-19-induced inequalities affect participants’ beliefs across the political divide? 4.3 Treatment effect by party identification To directly assess Hypotheses 2A and 2B, Table 2 reports average treatment effects and conditional treatment effects for the main models and three robustness checks (see Data and Methods). Conditional effects are estimated by interacting participants’ treatment condition and party identification.Table 2 Results from main model and alternative specifications. Table 2 Income Redistribution Vulnerable Measures As reported in manuscript (n = 1,003) Average treatment effect 0.26 *** 0.19 * 0.20 ** 0.15 Conditional treatment effects  Strong democrat 0.27 * 0.21 0.22 −0.13  Democrat 0.18 0.04 0.23 0.17  Neither 0.26 * 0.18 0.10 0.24  Republican 0.54 ** 0.72 ** 0.43 * 0.06  Strong republican 0.03 −0.38 0.11 0.87 * Check A (n = 1,003) Income Redistribution Vulnerable Measures Average treatment effect 0.28 *** 0.22 ** 0.21 ** 0.12 Conditional treatment effects  Strong democrat 0.29 * 0.32 * 0.24 * −0.09  Democrat 0.20 0.11 0.21 0.16  Neither 0.30 * 0.12 0.10 0.19  Republican 0.51 * 0.72 ** 0.44 * 0.03  Strong republican 0.06 −0.23 0.12 0.74 Check B (n = 1,003) Income Redistribution Vulnerable Measures Average treatment effect 0.28 *** 0.23 ** 0.21 ** 0.13 Conditional treatment effects  Strong democrat 0.29 * 0.32 * 0.24 * −0.09  Democrat 0.20 0.11 0.21 0.15  Neither 0.30 * 0.12 0.10 0.19  Republican 0.51 * 0.71 ** 0.44 * 0.03  Strong republican 0.07 −0.15 0.16 0.79 Check C (n = 937) Income Redistribution Vulnerable Measures Conditional treatment effects  Strong democrat 0.29 * 0.32 * 0.24 * −0.08  Democrat 0.19 0.10 0.22 0.16  Neither 0.36 ** 0.12 0.06 0.42 *  Republican 0.54 ** 0.71 ** 0.45 * 0.03  Strong republican 0.03 −0.27 0.10 0.77 Note. ‘Check A’ reports estimation results from OLS models including a bank of control variables; ‘Check B’ adds an attention check dummy to the controls; ‘Check C’ drops participants who picked “not applicable” in response to the party identification question. Income = “Differences in income in the United States are too large”; Redistribution = “It is the responsibility of the government to reduce the differences in income between people with high incomes and those with low incomes”; Vulnerable = “The most vulnerable in society are hit hardest by the coronavirus, also known as COVID-19”; Measures = “The measures taken against the coronavirus are more harmful than the virus itself.” *p < .05, **p < .01, ***p < .00 (two-tailed). To visualize our main findings, Fig. 3 plots predictive margins calculated from the regression models, comparing concerns about inequality and support for redistribution between participants in the treatment and control group by party identification. We find a significant difference between the controls and treated, indicative of a conditional treatment effect on participants’ belief that income inequality is too high, ranging from 0.28 points (95% CI, 0.03–0.51) among strong Democrats to 0.26 points for participants in the middle (95% CI, 0.07–0.53), and 0.54 points among Republicans (95% CI, 0.14–0.94). We also find positive differences between control and treatment group for Democrats (0.18 points) and strong Republicans (0.03 points), albeit not significantly different from zero at p < .05.Fig. 3 Beliefs about inequality by treatment and party identification. Note. Predictive margins based on OLS regression models including pretreatment attitudes and all control variables. Whiskers indicate the 95% confidence interval around each estimate. Fig. 3 We only find conditional treatment effects for participants’ support for government redistribution among Republicans, 0.72 points in size (95% CI, 0.21–1.22). Simply put, the informational treatment brings Republicans three-quarters up the way from “neither agree nor disagree” toward “somewhat agree.” These findings support the expectation that information provision impacts beliefs across partisan lines (Hypothesis 2B) and provide no support for the alternative expectation that the response to information is limited only to Democrats (Hypothesis 2A). Taken together, we find evidence of a shrinking ideological gap between moderate Republicans and Democrats following the provision of factual information about COVID-19's consequences for economic inequality. Among participants in the control group, the ideological gap in concerns about inequality is about 1 point between the partisan poles and 0.7 points when comparing moderate Republicans and Democrats. The partisan gap in support for redistribution is 0.9 and 1 point respectively. Our treatment reduces the gap between moderate Democrats and Republicans to 0.3 points on both topics (i.e. compared to 0.7 and 1); a gap that is not significantly different from zero at p < .05, reflective of an especially large attitudinal change among moderate Republicans. This finding however does not extend to those at the polar ends of the political spectrum. Comparing strong Democrats to strong Republicans in the treatment group, the polar ideological gap in concerns about inequality and support for redistribution increases to 1.2 and 1.5 points, respectively (compared to 1 and 0.9 in the control group). This finding adds an important qualification to our empirical support for Hypothesis 2B: provision of information affects beliefs on both sides of the political divide but may increase intraparty polarization. We return to this finding in our conclusion. Fig. 4 visualizes participants’ beliefs about COVID-19’s effects on the most vulnerable groups in society and the measures taken by government. For the former, we find a significant difference between the treated and controls, indicative of a conditional treatment effect of 0.43 points (95% CI, 0.07 - 0.85) among Republicans--almost half the way from “somewhat agree” to “agree”. For all other groups, we find positive differences between treated and controls, ranging from 0.10 points among participants in the middle, 0.23 for Democrats, 0.22 for strong Democrats, and 0.11 for strong Republicans, none of which however are significantly different from zero at p .05. Thus, we find a pattern of results comparable to those described above, lending no support for Hypothesis 2A and conditional support to Hypothesis 2B. Turning to participants’ beliefs about the negative consequences of government measures vis-à-vis the virus (Fig. 4 ), we find one statistically significant difference between the controls and treated: among strong Republicans, participants in the treatment group are more likely by 0.87 points (95% CI, 0.08–1.67) to believe that the measures taken by government are worse than the virus itself. Apparently, among these participants, 60 percent of which already believed the antidote to be worse than the illness (Fig. 1b), the informational treatment bolstered their concerns. This finding underlines the qualification we previously made with regard to the empirical support for Hypothesis 2B.Fig. 4 Beliefs about COVID-19 by treatment and party identification. Note. Predictive margins based on OLS regression models including pretreatment attitudes and all control variables. Whiskers indicate the 95% confidence interval around each estimate. Fig. 4 What does this mean for the ideological gap in beliefs about COVID-19-induced inequality and measures? Among control group participants, we find a partisan gap in the belief that COVID-19 has disproportionately affected the most vulnerable groups in society of 0.5 points both among moderates and strong partisans. The ideological gap in the belief that government measures have done more harm than the virus is 0.7 and 1.3 points, respectively. Our informational treatment reduces the gap in beliefs about COVID-19's consequences to 0.3 points between moderate Republicans and Democrats (not significantly different from zero at p < .05) but increases it to 0.6 points between the polar ends. Among moderates, concerning the role of government during the crisis we find a reduction in the attitudinal gap from 0.7 to 0.6 point (n.s.; p < .05). The gap between strong Democrats and strong Republicans however increases from 1.3 to 2.3, especially driven by the latter who become even more convinced that government COVID-19 measures have done more harm than the virus. We should note that this finding does not perfectly replicate in the robustness checks, which indicate similarly large coefficients for strong Republicans but report p-values just above the 5% level (Check A: 0.73, p = .07; Check B: 0.79, p = .06; Check C: 0.77, p = .06). 5 Conclusions Inspired by studies reporting surprisingly limited popular concern over steadily rising inequalities in recent decades (Breznau and Hommerich, 2019; Kenworthy and McCall, 2008; Luebker, 2014; Trump, 2017), this study asked whether a sudden and vast increase in inequalities in times of crisis (cf., Robinson et al., 2021) would make people's ideological beliefs more pliable than evidence from ‘normal’ times would suggest. Specifically, does the provision of factual information about COVID-19-induced inequalities make people more concerned and more supportive of income redistribution? We address this question by means of an original population-based survey experiment fielded in the least-likely case of the United States, renowned for its relatively high tolerance for income differences (Alesina et al., 2018) and partisan filtering of information (Bolsen et al., 2014). We find that Americans are split by partisan lines on each topic related to inequality. However, when exposed to the informational treatment, Americans across the ideological spectrum express 1) more concerns over economic inequality, 2) stronger support for income redistribution, and 3) stronger acknowledgement that the coronavirus has especially hurt the most vulnerable in society. These findings provide no support for the expectation that information about COVID-19-induced inequality only leads to belief change among Democrats. Instead, they lend support to the alternative Hypothesis that the provision of information in times of crisis may lead to belief change on both sides of the political divide. Research on the Great Recession in the United States (Margalit, 2013) and the Netherlands (Gidron and Mijs, 2019) documents growing support for redistribution especially among those who personally experienced economic hardship, reflecting self-interested concerns. Our findings suggest that a crisis like COVID-19 may fuel more broadly shared sociotropic concerns. Hence, further research into the specific conditions fueling both types of concern across the Atlantic would be most worthwhile. Turning to conditional treatment effects along ideological lines, in line with Bisgaard's study on the United Kingdom (2015) we find that even in the United States, moderate partisans are not far apart, assured they are provided with the same set of factual information. In the American case, the political divide is bridged by moderate Republicans who become markedly more supportive of redistribution following the informational treatment. However, while our informational treatment rendered non-significant the gap between moderates, it substantially increased the gap between moderate and strong Republicans. Moderate Republicans are receptive to new information and perspectives in times of crisis (cf. Swidler, 1986), whereas strong Republicans hold on to their political commitment or double down (cf. Nyhan and Reifler, 2010). We can think of three reasons for this intra-party polarization among Republicans. Perhaps our informational intervention, framed as non-partisan academic facts, rubbed strong Republicans the wrong way, given their relatively high levels of distrust of experts and the nonpartisan news media (Evans and Hargittai, 2020; Shepherd et al., 2020), in which this type of reporting is the default. Alternatively, strong Republicans could have attributed the steep rise in COVID-19-induced unemployment presented in the experimental condition as the consequence of government interventions, especially government shutdowns, hampering people from getting back to work. Such views typically go together both with their partisan position and demographic profile (as in other population-based surveys, strong Republicans in our sample were more likely than both moderate Republicans and all others to be male, white, Protestant, and have higher incomes and levels of education; see SI, Table S6 for details). Yet another factor may be that strong partisans -- Democrats and Republicans -- are simply more resistant to information that opposes their world view. Either way, we consider it less likely that information describing COVID-19-induced inequalities would invoke such polarization in other Western settings, as right-wing populist constituencies have been observed to be more pro-redistribution than those who prefer non-populist right-wing parties (De Koster et al., 2013). Given the ideological profile of the contemporary European left and right, information on cultural issues such as immigration and national sovereignty is more likely to incite polarization (Rydgren, 2008; Van Elsas et al., 2016). We welcome future research on the polarizing or unifying effects of informational treatments about inequalities in other settings. An important question that remains, concerns the scope and implications of our findings beyond the experimental setting. As the COVID-19 pandemic “does not appear to have fundamentally changed how subjects respond to treatments” in online experiments (Peyton et al., 2021), we are confident that exposure to information depicting the sudden and substantial increase in COVID-19-related inequality increases concerns among the bulk of the American population. The effect of information is likely to be greater in contexts with a lower tolerance of economic inequality. Yet, while our study exposes the public to the same set of information, in their daily lives, many Americans are exposed to partisan media mirroring their ideological profile and reinforcing their perspective (Bruine de Bruin et al., 2020). It follows that in contexts with a less polarized media landscape – such as the public broadcasting systems of various European countries (Mosca and Quaranta, 2016) – the public may be more uniformly informed about economic inequality, and consequently more likely to express concerns when confronted with information about its rampant rise. These suggest testable hypotheses for future research. Having established that most people – even the divided American public – express more concern about inequality when confronted with factual information, a subsequent question regards whether this finding translates to other contentious issues. For instance, can informational interventions produce a similar effect when targeting compliance with measures to mitigate the pandemic (see also Kelley and Evans, 2021)? Providing such information is challenging in a context where “[p]olitical leaders and media outlets on the right and left have sent divergent messages about the severity of the crisis, which could impact the extent to which Republicans and Democrats engage in social distancing and other efforts to reduce disease transmission” (Allcott et al., 2020, p. 1). Yet when information on measures to mitigate disease transmission is more uniformly available across the political landscape, there is good reason to believe the partisan gap can be bridged. To this point, Druckman et al. (2021, p. 36) observed that closing the partisan gap on the use of masks followed “changing rhetoric by Republican elites—including President Trump—to follow the Democratic perspective on mask wearing.” All in all, whereas steadily rising inequalities have sparked remarkably little public concern in recent decades, uniformly confronting the public with factual information describing the economic consequences of COVID-19 makes even the polarized American public more worried about inequality and more supportive of income redistribution. The strikingly uniform pattern of responses among moderates on both sides of the political divide suggest that disagreement over inequality may be rooted not in fundamentally incompatible worldviews but in different perceptions of how things are, which prove pliable through the provision of information. At the same time, our findings suggest that a crisis like COVID-19 may upend ideological rifts among moderates, while increasing their salience on the right of the political spectrum. We expect information describing a sudden and substantial increase in inequalities to be an even more likely source of concern in many European countries, where it is less likely to inspire attitudinal polarization at the political fringe. Funding Jonathan Mijs received funding from a 10.13039/100010665 Marie Skłodowska-Curie Individual Fellowship, EU Commission Horizon 2020 Grant no. 88296 and a Veni grant from the 10.13039/501100003246 Nederlandse Organisatie voor Wetenschappelijk Onderzoek (NWO), grant no. VI.Veni.201S.003. Willem de Koster and Jeroen van der Waal received financial support through Vidi grants from the 10.13039/501100003246 NWO , nos. 016.Vidi.185.207 and 452-17-009. Data availability Replication data and Stata scripts have been made available through the Open Science Framework to allow for independent verification of our findings: https://osf.io/ub538. Appendix A Supplementary data The following are the Supplementary data to this article:Multimedia component 1 Multimedia component 1 Multimedia component 2 Multimedia component 2 Appendix A Supplementary data to this article can be found online at https://doi.org/10.1016/j.ssresearch.2021.102692. ==== Refs References Alesina A. Stantcheva S. Teso E. Intergenerational mobility and preferences for redistribution Am. Econ. Rev. 108 2018 521 554 10.1257/aer.20162015 Allcott H. Boxell L. Conway J. Gentzkow M. Thaler M. Yang D. Polarization and public health: partisan differences in social distancing during the coronavirus pandemic J. Publ. Econ. 191 2020 104254 Aronow P.M. Baron J. Pinson L. A note on dropping experimental subjects who fail a manipulation check Polit. Anal. 27 2019 572 589 Athreya K. Mather R. Mustre-del-Rio J. Sánchez J.M. Employment Vulnerability and (Financial) Pre-existing Conditions 2020 Federal Reserve Bank of Richmond Atkinson A.B. Piketty T. Saez E. Top incomes in the long run of history J. Econ. Lit. 49 2011 3 71 10.1257/jel.49.1.3 Beall A.T. Hofer M.K. Schaller M. Infections and elections: did an Ebola outbreak influence the 2014 U.S. Federal elections (and if so, how)? Psychol. Sci. 27 2016 595 605 10.1177/0956797616628861 26976083 Berrebi C. Klor E.F. Are voters sensitive to terrorism? Direct evidence from the Israeli electorate Am. Polit. Sci. Rev. 102 2008 279 301 10.2307/27644521 Bird R. Ritter Z. Is the Media Creating Division on COVID-19 Health Practices? 2020 Gallup Washington, DC Bisgaard M. Bias will find a way: economic perceptions, attributions of blame, and partisan-motivated reasoning during crisis J. Polit. 77 2015 849 860 10.1086/681591 Bolsen T. Druckman J.N. Cook F.L. The influence of partisan motivated reasoning on public opinion Polit. Behav. 36 2014 235 262 10.1007/s11109-013-9238-0 Bradley D. Huber E. Moller S. Nielsen F. Stephens J.D. Distribution and redistribution in postindustrial democracies World Polit. 55 2003 193 228 Breznau N. Hommerich C. No generalizable effect of income inequality on public support for governmental redistribution among rich democracies 1987–2010 Soc. Sci. Res. 81 2019 170 191 10.1016/j.ssresearch.2019.03.013 31130195 Bruine de Bruin W. Saw H.-W. Goldman D.P. Political polarization in US residents' COVID-19 risk perceptions, policy preferences, and protective behaviors J. Risk Uncertain. 61 2020 177 194 10.1007/s11166-020-09336-3 33223612 Bureau of Labor Statistics Labor Force Statistics from the Current Population Survey 2020 U.S. Bureau of Labor Statistics Washington, DC Cappelen A.W. Falch R. Sorensen E.O. Tungodden B. Wezerek G. Opinion | what Do You Owe Your Neighbor? the Pandemic Might Change Your Answer 2020 N. Y. Times CBO The Distribution of Household Income, 2016 | Congressional Budget Office 2019 Congressional Budget Office Washington, DC Civiqs Coronavirus: Outbreak Concern 2020 Civiqs Oakland, CA Clifford S. Sheagley G. Piston S. Increasing precision without altering treatment effects: repeated measures designs in survey experiments Am. Polit. Sci. Rev. 2021 10.1017/S0003055421000241 (in press) Collins C. U.S. Billionaires got 62 percent richer during pandemic They’re Now up $1.8 Trillion 2021 Institute for Policy Studies Conway L.G. III Woodard S.R. Zubrod A. Chan L. Why are conservatives less concerned about the coronavirus (COVID-19) than liberals? Comparing political, experiential, and partisan messaging explanations Pers. Indiv. Differ. 183 2021 111 124 Cox J. The Upcoming Job Losses Will Be unlike Anything the US Has Ever Seen 2020 CNBC Cox R.H. The social construction of an imperative: why welfare reform happened in Denmark and The Netherlands but not in Germany World Polit. 53 2001 463 498 17595731 Dalton R.J. The quantity and the quality of party systems: party system polarization, its measurement, and its consequences Comp. Polit. Stud. 41 2008 899 920 10.1177/0010414008315860 De Koster W. Achterberg P. Van der Waal J. The new right and the welfare state: the electoral relevance of welfare chauvinism and welfare populism in The Netherlands Int. Polit. Sci. Rev. 34 2013 3 20 DeRensis H. The coronavirus economic crisis: the next Great depression? Natl 2020 Interest March 20 Druckman J.N. Klar S. Krupnikov Y. Levendusky M. Ryan J.B. Affective polarization, local contexts and public opinion in America Nat. Hum. Behav. 5 2021 28 38 10.1038/s41562-020-01012-5 33230283 Eadeh F.R. Chang K.K. Can threat increase support for liberalism? New insights into the relationship between threat and political attitudes Soc. Psychol. Personal. Sci. 11 2020 88 96 10.1177/1948550618815919 Evans J.H. Hargittai E. Who doesn't trust fauci? The public's belief in the expertise and shared values of scientists in the COVID-19 pandemic Socius 6 2020 10.1177/2378023120947337 2378023120947337 Flaxman S. Goel S. Rao J.M. Filter bubbles, echo chambers, and online news consumption Publ. Opin. Q. 80 2016 298 320 10.1093/poq/nfw006 Gerring J. Is there a (viable) crucial-case method? Comp. Polit. Stud. 40 2007 231 253 Gidron N. Mijs J.J.B. Do changes in material circumstances drive support for populist radical parties? Panel data evidence from The Netherlands during the Great recession, 2007–2015 Eur. Socio Rev. 2019 10.1093/esr/jcz023 Gollwitzer A. Martel C. Brady W.J. Pärnamets P. Freedman I.G. Knowles E.D. Van Bavel J.J. Partisan differences in physical distancing are linked to health outcomes during the COVID-19 pandemic Nat. Hum. Behav. 4 2020 1186 1197 33139897 Gordon R.J. The Rise and Fall of American Growth 2016 Princeton University Press Princeton, NJ Grasso M. Klicperová-Baker M. Koos S. Kosyakova Y. Petrillo A. Vlase I. The impact of the coronavirus crisis on European societies. What have we learnt and where do we go from here? – introduction to the COVID volume Eur. Soc. 2021 1 31 10.1080/14616696.2020.1869283 (in press) Grossman G. Kim S. Rexer J.M. Thirumurthy H. Political partisanship influences behavioral responses to governors' recommendations for COVID-19 prevention in the United States Proc. Natl. Acad. Sci. Unit. States Am. 117 2020 24144 24153 10.1073/pnas.2007835117 Havey N.F. Partisan public health: how does political ideology influence support for COVID-19 related misinformation? J. Comput. Soc. Sci. 3 2020 319 342 33163686 ISSP Research Group International Social Survey Programme 2019 Social Inequality V. Final Questionnaire 2018 GESIS Cologne Jerit J. Zhao Y. Political misinformation Annu. Rev. Polit. Sci. 23 2020 77 94 10.1146/annurev-polisci-050718-032814 Jost J.T. Glaser J. Kruglanski A.W. Sulloway F.J. Political conservatism as motivated social cognition Psychol. Bull. 129 2003 339 375 10.1037/0033-2909.129.3.339 12784934 Kelley J. Evans M.D.R. Legitimate earnings inequality and national welfare commitment: correspondence between economic institutions and the pay 80,000+ people in 30 nations think legitimate for ordinary jobs and for elite jobs Soc. Sci. Res. 94 2021 102446 10.1016/j.ssresearch.2020.102446 33648680 Kelly N.J. Enns P.K. Inequality and the dynamics of public opinion: the self-reinforcing link between economic inequality and mass preferences Am. J. Polit. Sci. 54 2010 855 870 10.1111/j.1540-5907.2010.00472.x Kenworthy L. McCall L. Inequality, public opinion and redistribution Soc. Econ. Rev. 6 2008 35 68 10.1093/ser/mwm006 Kenworthy L. Pontusson J. Rising inequality and the politics of redistribution in affluent countries Perspect. Polit. 3 2005 449 471 Kozlowski A.C. Murphy J.P. Issue alignment and partisanship in the American public: revisiting the ‘partisans without constraint’ thesis Soc. Sci. Res. 94 2021 102498 10.1016/j.ssresearch.2020.102498 33648681 Kreps S.E. Kriner D.L. Model uncertainty, political contestation, and public trust in science: evidence from the COVID-19 pandemic Sci. Adv. 6 2020 eabd4563 10.1126/sciadv.abd4563 32978142 Kuklinski J.H. Quirk P.J. Jerit J. Schwieder D. Rich R.F. Misinformation and the currency of democratic citizenship J. Polit. 62 2000 790 816 10.1111/0022-3816.00033 Kuziemko I. Norton M.I. Saez E. Stantcheva S. How elastic are preferences for redistribution? Evidence from randomized survey experiments Am. Econ. Rev. 105 2015 1478 1508 10.1257/aer.20130360 Lee K.S. Fujita Y. Economic recession and the nature and pace of social change in Japan Soc. Sci. Res. 40 2011 784 795 10.1016/j.ssresearch.2011.01.002 Lin W. Agnostic notes on regression adjustments to experimental data: reexamining Freedman's critique Ann. Appl. Stat. 7 2013 295 318 10.1214/12-AOAS583 Loveless M. Whitefield S. Being unequal and seeing inequality: explaining the political significance of social inequality in new market democracies Eur. J. Polit. Res. 50 2011 239 266 10.1111/j.1475-6765.2010.01929.x Luebker M. Income inequality, redistribution, and poverty: contrasting rational choice and behavioral perspectives Rev. Income Wealth 60 2014 133 154 10.1111/roiw.12100 Margalit Y. Explaining social policy preferences: evidence from the Great recession Am. Polit. Sci. Rev. 107 2013 80 103 10.1017/S0003055412000603 McCall L. Burk D. Laperrière M. Richeson J.A. Exposure to rising inequality shapes Americans' opportunity beliefs and policy support Proc. Natl. Acad. Sci. Unit. States Am. 114 2017 9593 9598 10.1073/pnas.1706253114 Mijs J.J.B. Visualizing Belief in Meritocracy, 1930-2010 2018 Socius 4 Mijs J.J.B. The paradox of inequality: income inequality and belief in meritocracy go hand in hand Socio-Economic Review 19 1 2021 7 35 10.1093/ser/mwy051 Mora G.C. Schickler E. Paschel T. Perceptions of Inequality and the Pandemic Vary Drastically Among Californians 2020 Inst. Gov. Stud. Release 2020-08 Mosca L. Quaranta M. News diets, social media use and non-institutional participation in three communication ecologies: comparing Germany, Italy and the UK Inf. Commun. Soc. 19 2016 325 345 Mummolo J. News from the other side: how topic relevance limits the prevalence of partisan selective exposure J. Polit. 78 2016 763 773 10.1086/685584 Naumann E. Buss C. Bähr J. How unemployment experience affects support for the welfare state: a real panel approach Eur. Socio Rev. 32 2016 81 92 Neimanns E. Busemeyer M.R. Garritzmann J.L. How popular are social investment policies really? Evidence from a survey experiment in eight western European countries Eur. Socio Rev. 34 2018 238 253 10.1093/esr/jcy008 Nyhan B. Reifler J. When corrections fail: the persistence of political misperceptions Polit. Behav. 32 2010 303 330 10.1007/s11109-010-9112-2 Oude Groeniger J. Noordzij K. van der Waal J. de Koster W. Dutch COVID-19 lockdown measures increased trust in government and trust in science: a difference-in-differences analysis Soc. Sci. Med. 275 2021 113819 10.1016/j.socscimed.2021.113819 33725488 Packer G. America's plastic hour is upon us Atlantic 2020 48 57 Palan S. Schitter C. Prolific.ac—a subject pool for online experiments J. Behav. Exp. Finance 17 2018 22 27 10.1016/j.jbef.2017.12.004 Pechar E. Bernauer T. Mayer F. Beyond political ideology: the impact of attitudes towards government and corporations on trust in science Sci. Commun. 40 2018 291 313 10.1177/1075547018763970 Peyton K. Huber G.A. Coppock A. The generalizability of online experiments conducted during the COVID-19 pandemic J. Exp. Polit. Sci. 2021 10.1017/XPS.2021.17 Robinson L. Schulz J. Ball C. Chiaraluce C. Dodel M. Francis J. Huang K.-T. Johnston E. Khilnani A. Kleinmann O. Cascading crises: society in the age of COVID-19 Am. Behav. Sci. 2021 00027642211003156 Rydgren J. Immigration sceptics, xenophobes or racists? Radical right-wing voting in six West European countries Eur. J. Polit. Res. 47 2008 737 765 Sachweh P. Sthamer E. Why do the affluent find inequality increasingly unjust? Changing inequality and justice perceptions in Germany, 1994–2014 Eur. Socio Rev. 35 2019 651 668 10.1093/esr/jcz024 Schüller S. The 9/11 conservative shift Econ. Lett. 135 2015 80 84 10.1016/j.econlet.2015.07.031 Schwartz N.D. ‘Nowhere to Hide’ as Unemployment Permeates the Economy 2020 N. Y. Times Shepherd H. MacKendrick N. Mora G.C. Pandemic politics: political worldviews and COVID-19 beliefs and practices in an unsettled time Socius 6 2020 10.1177/2378023120972575 2378023120972575 Suhay E. Klasnja M. Rivero G. Ideology of affluence: rich Americans' explanations for inequality and attitudes toward redistribution J. Polit. 2020 10.1086/709672 Swidler A. Culture in action: symbols and strategies Am. Socio. Rev. 51 1986 273 286 10.2307/2095521 Tierney K.J. From the margins to the mainstream? Disaster research at the crossroads Annu. Rev. Sociol. 33 2007 503 525 10.1146/annurev.soc.33.040406.131743 Trump K.-S. Income inequality influences perceptions of legitimate income differences Br. J. Polit. Sci. 48 2017 929 952 10.1017/S0007123416000326 Van der Waal J. Achterberg P. Houtman D. Class is not dead—it has been buried alive: class voting and cultural voting in postwar western societies (1956–1990) Polit. Soc. 35 2007 403 426 Van Elsas E.J. Hakhverdian A. Van der Brug W. United against a common foe? The nature and origins of Euroscepticism among left-wing and right-wing citizens W. Eur. Polit. 39 2016 1181 1204 Wagner‐Pacifici R. Theorizing the restlessness of events Am. J. Sociol. 115 2010 1351 1386 10.1086/651299 Waterfield S. A list of essential workers that we should thank and support during the coronavirus pandemic Newsweek 2020

34629159

PMC9754068

NO-CC CODE

2022-12-16 23:26:26

no

Placenta. 2021 Dec 28; 116:1

latin-1

Placenta

10.1016/j.placenta.2021.09.020

oa_other

==== Front J Thorac Cardiovasc Surg J Thorac Cardiovasc Surg The Journal of Thoracic and Cardiovascular Surgery 0022-5223 1097-685X Published by Elsevier Inc. on behalf of The American Association for Thoracic Surgery S0022-5223(20)32850-6 10.1016/j.jtcvs.2020.10.037 Commentary Commentary: From Locke to Merton to lung allocation: Unintended consequences Bremner Ross M. MD, PhD ∗ Norton Thoracic Institute, St Joseph's Hospital and Medical Center, Phoenix, Ariz ∗ Address for reprints: Ross M. Bremner, MD, PhD, Norton Thoracic Institute, St Joseph's Hospital and Medical Center, 500 W Thomas Rd, Suite 500, Phoenix, AZ 85013. 17 10 2020 1 2022 17 10 2020 163 1 348349 9 10 2020 9 10 2020 9 10 2020 © 2020 Published by Elsevier Inc. on behalf of The American Association for Thoracic Surgery. 2020 The American Association for Thoracic Surgery Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. ==== Body pmc Ross M. Bremner, MD, PhD Central Message The 250-mile geographic allocation rule may have unintended consequences. See Article page 339. The Law of Unintended Consequences: an intervention in a complex system tends to create unanticipated and often undesirable outcomes. —Wikipedia Allocation of the limited supply of donor organs has always been a great challenge in transplantation. Organ procurement agencies are responsible for the allocation of lungs according to the Final Rule1 passed in 1998 and enacted in 2000. The goal of this rule is to allocate organs equitably across the country. Before 2005, patients accrued time on a waiting list and then were offered organs according to the proximity of their “region” to the donor hospital. The change to the lung allocation scoring system in 2005 greatly improved upon the previous time accrual method and has resulted in many improvements, including far fewer patients dying on the wait list. However, organs were still allocated according to regions (areas assigned to the local organ procurement organizations—58 across the county). More recently, as a result of a lawsuit (and the urgent efforts of the United Network for Organ Sharing Thoracic Committee, over the Thanksgiving holiday, no less), the process of organ allocation with respect to regions was changed in 2017 to encompass a region within a 250 nautical-mile radius from the donor hospital, with a secondary offer to a region within a 500-mile radius. This is beautifully described by Alexandra Glazier,2 and I encourage you to read her article. The goal of this change was to more equitably distribute organs from a geographic standpoint. The effect of the rule has not been identical on all programs, as the authors of this manuscript point out.3 Haywood and colleagues3 have shown a fairly profound effect on their small- to medium-sized program, especially with an increase in the distance now traveled to get a donor organ, an increase in retrieval costs, and unfortunately, an increase in their wait-list mortality. Clearly, these are unintended consequences of this new geographic rule that are contrary to the goals of the Final Rule. The Discussion of their findings is important, especially as they raise the question about access to transplantation for those with lower socioeconomic status. However, their experience—and the experience of others—has not gone unnoticed. The Organ Procurement and Transplant Network recognizes that organ allocation is challenging and that our current system is not perfect. They are again working hard to improve these allocation systems. Currently, the Organ Procurement and Transplant Network is working on a new allocation system called continuous distribution.4 This new scoring system will take into account (1) medical urgency, (2) post-transplant survival, (3) candidate biology, (4) patient access, and (5) placement efficiency. Complex algorithms will be used, and theoretical modeling and analysis will be performed to “help identify any potential unintended consequences.” As a surgeon who has been in the lung transplant world since before the turn of the century (ouch), I have experienced the continued progress that has been made with respect to lung allocation. This is hard work and I laud all those involved in improving our allocation systems. I suspect I will witness many further improvements before the end of my career. Disclosures: The author reported no conflicts of interest. The Journal policy requires editors and reviewers to disclose conflicts of interest and to decline handling or reviewing manuscripts for which they may have a conflict of interest. The editors and reviewers of this article have no conflicts of interest. ==== Refs References 1 U.S. Department of Health and Human Services Organ procurement and transplantation network; final rule Fed Regist 42 CFR part 121 1999 56649 566616 2 Glazier A. The lung lawsuit: a case study in organ allocation policy and administrative law J Health Biomed L XIV 2018 139 148 3 Haywood N. Mehaffey J.H. Kilbourne S. Mannem H. Weder M. Lau C. Influence of broader geographic allograft sharing on outcomes and cost in smaller lung transplant centers J Thorac Cardiovasc Surg 163 2022 339 345 33008575 4 U.S. Department of Health and Human Services Organ procurement and transplantation network; continuous distribution Available at: https://optn.Transplant.Hrsa.Gov/governance/policy-initiatives/continuous-distribution/

33172661

PMC9754069

NO-CC CODE

2022-12-16 23:26:26

no

J Thorac Cardiovasc Surg. 2022 Jan 17; 163(1):348-349

utf-8

J Thorac Cardiovasc Surg

10.1016/j.jtcvs.2020.10.037

oa_other

==== Front J Thorac Cardiovasc Surg J Thorac Cardiovasc Surg The Journal of Thoracic and Cardiovascular Surgery 0022-5223 1097-685X Published by Elsevier Inc. on behalf of The American Association for Thoracic Surgery S0022-5223(20)32710-0 10.1016/j.jtcvs.2020.09.105 Commentary Commentary: The cost of drawing the line: Lung transplant programs in the post-donor service area era Fernandez Ramiro MD Rappaport Jesse MD Ahmad Usman MD ∗ Department of Cardiothoracic Surgery, Heart and Vascular Institute, Cleveland Clinic, Cleveland, Ohio ∗ Address for reprints: Usman Ahmad, MD, Department of Cardiothoracic Surgery, Heart and Vascular Institute, Cleveland Clinic, Cleveland, OH 44195. 3 10 2020 1 2022 3 10 2020 163 1 350352 22 9 2020 22 9 2020 23 9 2020 © 2020 Published by Elsevier Inc. on behalf of The American Association for Thoracic Surgery. 2020 Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. ==== Body pmc Lung transplant programs in post-donor service area era. Central Message Broader sharing has increased travel distance and cost. National waitlist mortality has remained stable in the early period. Innovative strategies are needed to overcome the shortcomings of this model. See Article page 339. Almost 3 years have passed since the emergency change in lung allocation policy in November 2017 that replaced the donor service area (DSA) with a 250 nautical mile radius as the first unit of lung allocation. The Organ Procurement and Transplantation Network Thoracic Transplantation Committee has monitored outcomes very closely, most recently reporting on 2-year outcomes since the change.1 Meanwhile, individual transplant centers have begun reporting their experiences. While the overarching goal is improvement in patient outcomes, the effects may not be identical across all programs in the United States. While we wait for longer term outcomes to mature, early effects can be seen in changes in cost of doing a transplant and potentially waitlist mortality. Haywood and colleagues2 have shared the University of Virginia experience, presenting it as a low- to medium-volume transplant center. Not unexpectedly, they found a significant decrease in local donors (6% vs 68%) and corresponding increase in procurement travel distance (145 vs 235 miles). This was associated with a significant increase in procurement costs per transplant ($60,852 vs $69,052). The findings of decreased local donors and escalating costs under the new allocation policy are consistent with the report by Puri and colleagues3 describing the Washington University in St Louis experience. Remarkably, the Washington University in St Louis group reported a doubling of cost per procurement ($34,000 to $70,203) associated with a rise in distant procurements. This phenomenon is reflected nationally as well, with decreased local donors reflected in longer travel distances.1 Despite increased travel distance and greater recipient lung allocation scores, the University of Virginia group found no differences in short-term mortality or primary graft dysfunction incidence. Interestingly, the authors reported an increase in waitlist mortality (31.6 vs 6.9 per 100 patient-years) that has not been reported previously.1 , 3 Closer examination of the data reveals one waitlist death under the former allocation policy compared with 5 deaths under the new policy. Two of the 5 deaths under the new policy were in delisted patients, making their inclusion in the final sum debatable. Furthermore, when comparing small absolute figures, minor differences can have a large relative effect. It is important to note that the national data at the 2-year time point after the elimination of the DSA showed no difference in waitlist mortality.1 Broader sharing through the new allocation policy was intended to help the sickest patients and decrease waitlist mortality. While some high lung allocation score groups trended toward decreased waitlist mortality in the 2-year Organ Procurement and Transplantation Network report,1 whether and when the new policy lowers waitlist mortality overall remains to be seen. Broader geographic distribution of organs as part of the Final Rule is here to stay.4 For programs to remain aggressive about donor evaluation and organ procurement over longer distances, centers need greater financial flexibility. A dry run over a longer distance is going to cost more and increase the financial strain on the system. This may eventually decrease a center's propensity to evaluate marginal donors and thus defeat the main purpose of abrogation of the DSA. Therefore, cost-containment measures such as resource sharing may be necessary for transplant programs. Puri and colleagues3 suggested the use of local procurement teams to mitigate travel costs. This strategy might have been difficult to implement during normal times. The coronavirus disease 2019 pandemic, with its limitations on interstate travel, has put it effectively in action. In combination with wider use of ex vivo perfusion (EVLP) to allow evaluation of organs, programs could potentially save the cost of a dry run. Another key factor in increasing adoption of this paradigm is the mutual understanding between the transplant program and the Organ Procurement Organization to waive part or all of the cost if the organ does not end up being transplanted. Programs that do not have in-house EVLP facilities should strongly consider joining the national clinical trials. Geographic inequity in lung allocation has existed since lung transplantation became an established therapy for end-stage lung disease.5 , 6 In addition to broader sharing of a scare resource, increasing the supply can also effectively mitigate some of the competing risk. Expanding the use of donation after circulatory death lungs and EVLP to rehabilitate marginal lungs are well-established ways to increase the donor pool.7 , 8 The International Society of Heart and Lung Transplant 5-year report on donation after circulatory death lung transplants found similar 30-day, 1-year, and 5-year survival compared with brain dead donors, supporting the effectiveness of this underused donor pool.9 Similarly, the use of EVLP on high risk lungs that otherwise would have been discarded has shown favorable outcomes, with similar survival and rejection rates compared to standard criteria lungs over the long term.8 , 10 As we move forward in this new era of lung allocation, it is imperative we continually seek ways to improve outcomes in the current environment since any change to lung allocation policy takes time. Increasing the donor pool and developing cost containment measures are actionable steps to continually expand lung transplantation while keeping it financially sustainable in the post-DSA era. Disclosures: The authors reported no conflicts of interest. The Journal policy requires editors and reviewers to disclose conflicts of interest and to decline handling or reviewing manuscripts for which they may have a conflict of interest. The editors and reviewers of this article have no conflicts of interest. ==== Refs References 1 Committee OTT Monitoring of the Lung Allocation Change, 2 Year Report Removal of DSA as a Unit of Allocation 2020 Available at: https://optn.transplant.hrsa.gov/media/3661/item_25_thoracic_committee_20200212.pdf 2 Haywood N. Mehaffey J.H. Kilbourne S. Mannem N. Weder M. Lau C. Influence of broader geographic allograft sharing on outcomes and cost in smaller lung transplant centers J Thorac Cardiovasc Surg 163 2022 339 345 33008575 3 Puri V. Hachem R.R. Frye C.C. Harrison M.S. Semenkovich T.R. Lynch J.P. Unintended consequences of changes to lung allocation policy Am J Transplant 19 2019 2164 2167 30758137 4 Network OPaT Final Rule 2000 Available at: https://optn.transplant.hrsa.gov/governance/about-the-optn/final-rule/ 5 Benvenuto L.J. Anderson D.R. Kim H.P. Hook J.L. Shah L. Robbins H.Y. Geographic disparities in donor lung supply and lung transplant waitlist outcomes: a cohort study Am J Transplant 18 2018 1471 1480 29266733 6 Kosztowski M. Zhou S. Bush E. Higgins R.S. Segev D.L. Gentry S.E. Geographic disparities in lung transplant rates Am J Transplant 19 2019 1491 1497 30431704 7 Ahmad U. Commentary: lung donation after circulatory death in the United States. Current and future challenges J Thorac Cardiovasc Surg 161 2021 467 468 32427149 8 Cypel M. Yeung J.C. Liu M. Anraku M. Chen F. Karolak W. Normothermic ex vivo lung perfusion in clinical lung transplantation N Engl J Med 364 2011 1431 1440 21488765 9 Van Raemdonck D. Keshavjee S. Levvey B. Cherikh W.S. Snell G. Erasmus M. Donation after circulatory death in lung transplantation-five-year follow-up from ISHLT Registry J Heart Lung Transplant 38 2019 1235 1245 31777330 10 Divithotawela C. Cypel M. Martinu T. Singer L.G. Binnie M. Chow C.W. Long-term outcomes of lung transplant with ex vivo lung perfusion JAMA Surg 154 2019 1143 1150 31596484

33208258

PMC9754070

NO-CC CODE

2022-12-16 23:26:26

no

J Thorac Cardiovasc Surg. 2022 Jan 3; 163(1):350-352

utf-8

J Thorac Cardiovasc Surg

10.1016/j.jtcvs.2020.09.105

oa_other

==== Front J Thorac Cardiovasc Surg J Thorac Cardiovasc Surg The Journal of Thoracic and Cardiovascular Surgery 0022-5223 1097-685X by The American Association for Thoracic Surgery S0022-5223(20)32840-3 10.1016/j.jtcvs.2020.10.029 Commentary Commentary: All politics are local. The impact of the new organ allocation system on a lung transplant center Mason David P. MD ∗ Baylor University Medical Center, Dallas, Tex ∗ Address for reprints: David P. Mason, MD, Baylor University Medical Center, 3410 Worth St, Suite 545, Dallas, TX 75246. 17 10 2020 1 2022 17 10 2020 163 1 349350 8 10 2020 8 10 2020 9 10 2020 © 2020 by The American Association for Thoracic Surgery. 2020 The American Association for Thoracic Surgery Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. ==== Body pmc David P. Mason, MD Central Message Change from a donor service area system of lung distribution to one based on proximity to the donor hospital impacts individual transplant centers in unpredictable ways. See Article page 339. Haywood and colleagues1 at the University of Virginia lung transplant program describe in their manuscript “Impact of Broader Geographic Sharing on Outcomes and Cost in Smaller Lung Transplant Centers” the effect of national changes that were made in lung distribution and implemented on November 24, 2017. Before this time, lungs for potential adult recipients were allocated according to donation service area (DSA), with prioritization of allocation to patients in the local DSA according to their lung allocation score (LAS). The DSA system of distribution was felt to be unfair because lungs were being offered to less-sick patients in a local DSA whereas sicker patients, even in close proximity but in another DSA, were passed over. Theoretically, this system could be contributing to increased waitlist deaths. Hence, the DSA allocation system was replaced by a system whereby lungs were first allocated by LAS to adult recipients within a 250 mile radius around the donor hospital. When considering this new system, all transplant centers were asking themselves the exact questions that Haywood and his colleagues attempt to answer in this manuscript. How will these nationally implemented changes impact my program on a local level? More specific questions included the following: (1) Will the geographic location of our program disadvantage us? (2) Will we need to travel further to obtain lungs? (3) Will these changes lead to more deaths on our waitlist? (4) Will it cost us more money to transplant? The authors answered many of these questions at their own center, and some of the answers that are published here are concerning. The number of transplants that they performed in an equivalent time frame before and after implementation of the new system did increase. Travel distance, travel cost, and total procurement costs increased. Waitlist mortality increased. Of greatest concern, the waitlist mortalities they observed after conversion to the new system all occurred in patients with very high LAS scores, the very patients the new system was designed to help. The results of the authors' findings are difficult to generalize among other transplant centers. The authors argue that they are disadvantaged by the relatively small size of their transplant program, its rural location, and service of the socioeconomically disadvantaged. They also cite the challenges of competing with other larger transplant centers within a 250-mile radius. However, it seems that every transplant center, not just their own, struggles to deal with its own unique set of challenges presented by the new system. This observational study cannot control for the multitude of variables related to organ distribution and transplantation. These include program reputation, quality of local organ procurement agencies, cultural donation patterns, patient referral patterns, competing transplant programs, and local population size to name just a few. The 1-year report published by the Organ Procurement and Transplantation Network Thoracic Transplantation Committee looked at the early national results after removal of DSA as a unit of a lung allocation.2 Although there were some statistically significant differences between pre- and post-era metrics, the jury is still out on the overall impact of the change. More data, including 1-year post-transplant survival, will need to be analyzed. What is certain is that changes in national donation policy clearly impact each of our transplant programs individually and differently. Disclosures: The author reported no conflicts of interest. The Journal policy requires editors and reviewers to disclose conflicts of interest and to decline handling or reviewing manuscripts for which they may have a conflict of interest. The editors and reviewers of this article have no conflicts of interest. ==== Refs References 1 Haywood N. Mehaffey J.H. Kilbourne S. Mannem H. Weder M. Lau C. Impact of broader geographic allograft sharing on outcomes and cost in smaller lung transplant center J Thorac Cardiovasc Surg 163 2022 339 345 33008575 2 Monitoring of the lung allocation change, 1 year report Removal of DSA as a Unit of Allocation. OPTN Thoracic Transplantation Committee. Available at: https://optn.transplant.hrsa.gov/media/2815/20190116_thoracic_committee_report_lung.pdf Accessed September 15, 2020

33248678

PMC9754071

NO-CC CODE

2022-12-16 23:26:26

no

J Thorac Cardiovasc Surg. 2022 Jan 17; 163(1):349-350

utf-8

J Thorac Cardiovasc Surg

10.1016/j.jtcvs.2020.10.029

oa_other

==== Front Sci Total Environ Sci Total Environ The Science of the Total Environment 0048-9697 1879-1026 Elsevier B.V. S0048-9697(21)00292-8 10.1016/j.scitotenv.2021.145226 145226 Article Letter to the editor regarding Hospers et al. (2020): Electric fans: A potential stay-at-home cooling strategy during the COVID-19 pandemic this summer? Wang Faming ⁎ School of Architecture and Art, Central South University, Changsha, China ⁎ School of Architecture and Art, Central South University, Changsha 410083, China. 16 1 2021 15 5 2021 16 1 2021 769 145226145226 14 12 2020 13 1 2021 © 2021 Elsevier B.V. All rights reserved. 2021 Elsevier B.V. Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. Editor: Jay Gan ==== Body pmcIn Hospers and colleagues' recent article (Hospers et al., 2020), electric fans have been proposed as a potential stay-at-home cooling strategy during the COVID-19 pandemic under heat wave conditions. Besides, the authors defined the threshold temperatures for electric fan-use so that the public could use this as a guideline. In this letter, I would like to challenge the rigorousness of the methodology used in their work to determine threshold temperature and relative humidity (RH) zone for electric fan-use during heatwave temperatures. Hospers et al. (2020) used a standard conceptual human heat balance model to analyze the heat transfer between the body and ambient environments with and without an electric fan. If the convective heat load with fan-use exceeded the maximum evaporative cooling potential (which was calculated from the maximum sweat production of an individual person), electric fan-use under the specific is not advisable. The above methodology seems flawed because in some recommended climatic conditions which fan-use is advisable, fanning could accelerate human sweating and thereby hasten dehydration. Hence, it's debatable to include those conditions for fan-use. In this letter, a rational model, i.e., the predicted heat strain model (ISO 7933, 2004) was selected to examine the effect of fanning on temporal variations of thermophysiological responses at various temperatures from 36 to 50 °C. The model was first validated by documented human trials in literatures (Ravanelli et al., 2015; Morris et al., 2019). Also, a guideline on the fan use to bring remarkable benefits of body cooling was proposed by performing simulations under various relative humidity (RH) levels (range: 10–90%, elevation intervals: 5%) at temperatures of 36–50 °C. To determine the threshold temperature-RH zone for the fan-use, the two most important thermophysiological parameters, i.e., the predicted core temperature and the total sweating production, in Fan (i.e., with the electric fan) and No Fan were compared. Fan use in a specific condition is advisable if any of the following criteria were met.i) The use of electric fan should decrease the core temperature by at least 0.3 °C within 2 h on a standard lightly-clothed healthy man shown in Hospers et al.'s article (i.e., body mass 70 kg, height 1.73 m, body surface area 1.83 m2, clothing insulation 0.1291 m2·K·W−1, evaporative resistance 23.7 m2·kPa−1·W−1, metabolic rate 65 W·m−2) and the core temperature with fan-use should never exceed 38.0 ± 0.2 °C (safety threshold core temperature (WHO, 1969)); ii) The electric fan should bring remarkable sweating suppression, i.e., the total sweat production should be decreased by 87.5 g per hour (thirst sensation is triggered with a body water loss of about 1% (Saltmarsh, 2008), i.e., 700 g). Morris et al. (2019) examined the effect of fan use on thermal and cardiovascular strain, risk of dehydration and thermal comfort of 12 healthy men during the peak condition in two heat wave weather conditions (i.e., 40 and 47 °C). The rectal temperatures at the end of 2-h exposure at 40 °C & 50%RH were 37.4 and 37.5 °C in Fan and No Fan, respectively. Post-exposure rectal temperatures at 47 °C & 10%RH were 37.7 and 37.4 °C, respectively. Simulation results obtained in two example heatwave conditions (i.e., 40 and 47 °C) are listed in Table 1 . It is evident that simulation results shown in Table 1 are quite close to observed clinical trial data reported in Morris et al.'s (2019) work and the maximal difference on core temperatures was 0.2 °C. Further, the model was validated by the human trial data reported in Ravanelli et al. (2015), see Fig. 1 . The predicted core temperature changes in different RH levels showed perfect match with those observed on human participants with the maximum temperature difference of <0.10 °C.Table 1 Post-simulation rectal temperatures and sweat production at 40 & 47 °C under various RH levels. Table 1Relative humidity, RH (%) Temperature: 40 °C Temperature: 47 °C Core temperature (°C) Total sweat production (g) Core temperature (°C) Total sweat production (g) Fan No Fan Fan No Fan Fan No Fan Fan No Fan 10 37.4 37.2 690 560 37.6 37.2 1100 920 15 37.4 37.1 690 570 37.6 37.1 1110 970 20 37.4 37.1 690 580 37.6 37.0 1110 1070 25 37.4 37.1 690 590 37.5 37.4 1130 1230 30 37.4 37.1 690 610 37.5 38.1 1160 1250 35 37.3 37.1 680 640 37.4 38.8 1220 1250 40 37.3 37.0 690 680 37.8 39.6 1250 1250 45 37.3 37.0 690 780 38.6 40.4 1240 1240 50 37.3 37.4 700 930 39.8 41.2 1240 1240 55 37.3 37.9 720 1190 – – – – 60 37.2 38.5 780 1240 – – – – 65 37.1 39.0 1010 1240 – – – – 70 38.2 39.6 1200 1240 – – – – 75 39.5 40.1 1230 1240 – – – – 80 – – – – – – – – 85 – – – – – – – – 90 – – – – – – – – Note: –, conditions are rarely seen on the Earth. Hence, simulations were not performed under those conditions Baseline rectal temperature: 36.8 °C; model inputting variables: clothing insulation: 0.33 clo, permeability index: 0.38, metabolic rate: 70 W/m2 fanning velocity: 2.0 m·s−1, simulation duration: 2 h. Fig. 1 Core temperature changes reported in human trials conducted by Ravanelli et al.'s study (i.e., ΔTcore_exp) and those predicted by the PHS model (i.e., ΔTcore_sim). Fig. 1 Thus, the PHS model could be able to generate accurate thermophysiological data for electric fan use studies. Table 1 shows that fan use is advisable at 40 °C when RHs are within 50–70%. Interestingly, the total sweat production of Fan is also lower than those of No Fan. Fanning greatly prompts sweat evaporation, which brings greater evaporative body cooling compared to No Fan. Thereby, less sweating is required to maintain the body heat balance. Contrary to existing guideline (Gupta et al., 2012), fan cooling could still be effective for the public without access to HVAC in heat wave periods if RH values fall within 50–70% at 40 °C. Similarly, fans are advisable if RHs fall within 30–40% at 47 °C. The recommended RH zone for fanning to alleviate physiological strain on healthy individuals under other heat wave conditions (36–50 °C) is presented in Fig. 2 (Left). It should be noted that a fanning speed of 2.5 m/s was used in the simulations because most published literatures on electric fans had a maximum speed of about this value (Yang et al., 2015; Liu et al., 2018). Obviously, recommended fanning application RH values decrease with the increasing environmental temperature. The maximum advisable fan-use temperature is 48 °C (with RHs = 30–35%). Fanning is advisable when the environment water vapour pressure is within 3.1–5.1 kPa. For the oelderly people, the maximum advisable fan-use air temperature is 42 °C (with RHs = 45–50%).Fig. 2 Left: Recommended fanning zone for the healthy young people (the blue shaded region) and the elderly people (i.e., OLD, the interior region bounded by the closed red curve) under various heat wave conditions (fanning speed: 2.5 m·s−1); Right: Temporal variations of rectal temperatures in scenarios of No Fan and Fan at 40 °C and RH = 50% (or RH = 70%). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) Fig. 2 Gupta et al. (2012) suggested that fan use might lead to increased dehydration risk. Ravanelli et al. (2015) and Ravanelli and Jay (2016) confirmed fanning could hasten dehydration. However, above statements and findings seem only partly correct. For instance, at 40 °C and 10–40%RH or at 47 °C & 10–20%RH, fanning did induce greater sweat production than No Fan. However, when RH goes above those mentioned ranges, fanning didn't actually lead to greater sweating production compared to No Fan. Besides, Fig. 2 (Right) demonstrates that fan use elevated rectal temperature at the initial exposure stage, the duration of which depends on the RH level. This is because sweating is still in its development phase in the initial heat exposure period. Hence, convective body heat gain outweighs the evaporative cooling induced by fanning. With the increasing saturation on the skin surface, heat dissipation through sweat evaporation becomes greater than convective heat gain due to fanning. Thus, a body heat balance might still be maintained if fanning is used in certain heat wave conditions (e.g., 40 °C & 50%RH). It is clear from Fig. 1 presented in Hospers et al.'s study (Hospers et al., 2020) that the threshold temperature-RH zone for fanning should be revised and narrowed down because fanning under some recommended environmental conditions could not bring remarkable cooling effect to the occupants. Conversely, fanning could hasten dehydration in some recommended conditions. Besides, the core temperature could be increased due to fanning. In particular, Hospers et al. used a surprisingly high fanning speed, 4.5 m/s. The threshold temperature-RH zone for effective fanning should be further narrowed down because the convective heat load due to fan-use is much greater at a speed of 4.5 m/s compared to that at 2.5 m/s. In summary, electric fans might be served as effective and economical personal cooling devices to reduce heat strain during heat waves if environmental RH levels are appropriate. Fanning during heat waves with either too low or too high RHs could worsen human thermophysiological status on lightly-clothed healthy people. Therefore, caution should still be taken when choosing fanning as an effective and economical strategy to mitigate heat strain under various heat wave conditions. Declaration of competing interest None. ==== Refs References Gupta S. Carmichael C. Simpson C. Clarke M.J. Allen C. Gao Y. Chan E.Y.Y. Murray V. Electric fans for reducing adverse health impacts in heatwaves Cochrane Database Syst. Rev. 7 2012 CD009888 Hospers L. Smallcombe J.W. Morris N.B. Capon A. Jay O. Electric fans: a potential stay-at-home cooling strategy during the COVID-19 pandemic this summer? Sci. Total Environ. 747 2020 141180 ISO ISO 7933-Ergonomics of the Thermal Environment-Analytical Determination and Interpretation of Heat Stress Using Calculation of the Predicted Heat Strain 2004 International Organization for Standardization Geneva, Switzerland Liu S. Lipczynska A. Schiavon S. Arens E. Detailed experimental investigation of air speed field induced by ceiling fans Build. Environ. 142 2018 342 360 Morris N.B. Hospers L. Capon A. Jay O. The effects of electric fan use under differing resting heat index conditions: a clinical trial Ann. Intern. Med. 171 2019 675 677 31382270 Ravanelli N.M. Jay O. Electric fan use in heat waves: turn on or turn off? Temperature 3 2016 358 360 Ravanelli N.M. Hodder S.G. Havenith G. Jay O. Heart rate and body temperature responses to extreme heat and humidity with and without electric fans J. Am. Med. Assoc. 313 2015 724 725 Saltmarsh M. Thirst: or, why do people drink? Nutr. Bull. 26 2008 53 58 WHO Health Factors Involved in Working under Conditions of Heat Stress. Technical Report No. 412 1969 World Health Organization Geneva, Switzerland Yang B. Schiavon S. Sekhar C. Cheong D. Tham K.W. Nazaroff W.W. Cooling efficiency of a brushless direct current stand fan Build. Environ. 85 2015 196 204

33482553

PMC9754072

NO-CC CODE

2022-12-16 23:26:26

no

Sci Total Environ. 2021 May 15; 769:145226

utf-8

Sci Total Environ

10.1016/j.scitotenv.2021.145226

oa_other

==== Front J Thorac Cardiovasc Surg J Thorac Cardiovasc Surg The Journal of Thoracic and Cardiovascular Surgery 0022-5223 1097-685X by The American Association for Thoracic Surgery S0022-5223(20)32896-8 10.1016/j.jtcvs.2020.10.072 Commentary Commentary: The 250-mile radius rule in lung transplant donation: Even the best intentions have untoward consequences Loor Gabriel MD ab∗ Mattar Aladdein MD a a Division of Cardiothoracic Transplantation and Circulatory Support, Baylor College, of Medicine, Houston, Tex b Division of Cardiothoracic Transplantation and Circulatory Support, Texas Heart Institute, Houston, Tex ∗ Address for reprints: Gabriel Loor, MD, Division of Cardiothoracic Transplantation and Circulatory Support, Texas Heart Institute, 6770 Bertner Ave, Suite, C-355K, Houston, TX 77030. 31 10 2020 1 2022 31 10 2020 163 1 346347 9 10 2020 9 10 2020 17 10 2020 © 2020 by The American Association for Thoracic Surgery. 2020 The American Association for Thoracic Surgery Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. ==== Body pmc Gabriel Loor, MD, and Aladdein Mattar, MD Central Message In a retrospective single-center study representing a moderate-size lung transplant program, the 250–mile-radius rule did not translate into greater donor organ utilization or waitlist survival. See Article page 339. The United Network for Organ Sharing (UNOS) sought to promote the fair and appropriate distribution of donor lungs by changing its allocation system to prioritize recipients within a 250–nautical-mile radius of the donor hospital.1 This change was made in response to a lawsuit over a recipient who wanted broader organ sharing. Previously, allocation had been based on the donor service area, such that a recipient's center would preferentially receive offers from a donor hospital that belonged to the donor service area, regardless of the distance between them. When UNOS changed this policy in 2017, the effects of this change were almost impossible to predict because of the unequal distribution of donor hospitals throughout the United States. Some transplant centers have many donor hospitals nearby and few recipient centers, whereas others have few donor hospitals and several recipient centers. Some centers are located near the coastline, which reduces their donor offers even further. Haywood and colleagues2 from the University of Virginia describe their real-world experience with the untoward effects of the 250-mile rule on small-to-moderate–sized programs. Such programs are responsible for nearly half of the lung transplants performed in the United States. The authors provide data on their program, which services central and southwestern Virginia. Their 250-mile region is saturated with 8 other lung transplant centers. The authors noted several effects associated with the change in the allocation system. Lung transplant volume increased 15%. But at the same time, the number of local donors decreased 11-fold, organ travel time increased 62%, and procurement costs increased 13%. Most importantly, waitlist mortality increased 4.5-fold. There was no increase in average total ischemic time, or in the rate of primary graft dysfunction or short-term mortality. As in any retrospective study, it is always possible that other confounders were responsible for the findings. It is difficult to reconcile the increased transplant volume with the increased waitlist deaths. Typically, a center whose volume increases will see fewer deaths on the waitlist. In this series, patients who died on the waitlist had high lung allocation scores (>80), which is a known risk factor.3 The authors also saw a 30% increase in listings after the policy change. Thus, it is conceivable that the increase in deaths on the waitlist was due not only to the allocation change but also to an increased listing of sicker recipients. Although the authors emphasize the increased costs, one could assume that the revenue from the increased transplant numbers could have made up for these costs. Nonetheless, the findings show reasonable evidence of increased travel times and costs associated with the new policy. This, in combination with the increased waitlist mortality, provides sobering evidence that this policy change may have had the unintended consequence of decreasing organ availability and reducing the fairness of organ allocation. It suggests that small-to-moderate–size programs may be unfairly influenced. Larger programs have a tendency to transport organs across longer distances and accept greater numbers of extended criteria donors. They build systems around these practices and implement backup procurement teams, standby aircraft, and ex vivo lung perfusion (EVLP) technology. This raises the issue of regionalization. Perhaps it is time that centers combine their strengths and reduce unnecessary competition that could adversely influence recipients. Unfortunately, such collaboration is difficult in our current health care system, in which even 2 transplant centers on the same block have different medical leadership boards, insurance contracts, and medical record systems, among other differences. Another way to promote the fair and appropriate distribution of donor organs would be to increase the radius to 500 miles. This was not done by UNOS because of concerns over cold ischemic time. Although some authors have suggested that cold ischemic intervals can be safely extended beyond 6 hours, this notion is not universally accepted.4 However, even with the increased mile limit, a donor organ can be transplanted with an ischemic time <6 hours. Haywood and colleagues2 show that total ischemic time was kept the same even when donor organs traveled nearly twice the distance. This is probably because of the speed of air travel and, potentially, intraoperative procedural changes, which were not specified. Portable normothermic EVLP with the Organ Care System Lung also reduces ischemia by perfusing and ventilating the lung throughout transportation.5 This benefit comes at a significantly increased procurement cost. Moreover, some regional perfusion centers receive an organ on ice, place it in static EVLP, and ship it back to a center for transplantation. This practice is based on studies showing no increase in adverse events with longer cold intervals and static EVLP.6 With present technology and practice patterns, increasing the radius to 500 nautical miles is probably a reasonable way to improve access to donor organs. The study by Haywood and colleagues2 raises concern about untoward consequences of the new allocation system, which has potentially increased waitlist deaths and raised procurement costs at small-to-moderate–size centers. Policy changes in solid-organ transplantation should be considered carefully before implementation and then constantly reviewed to examine their effects. Despite the best of intentions, such changes may not always achieve what they are meant to. Stephen N. Palmer, PhD, ELS, contributed to the editing of the manuscript. Disclosures: Dr Loor is a consultant for Abiomed and his institution receives grant support for extracorporeal membrane oxygenation and cardiothoracic transplant research from Maquet, 10.13039/100004374 Medtronic , Transmedics, and 10.13039/100006279 St. Jude Medical . Dr Mattar reported no conflicts of interest. The Journal policy requires editors and reviewers to disclose conflicts of interest and to decline handling or reviewing manuscripts for which they may have a conflict of interest. The editors and reviewers of this article have no conflicts of interest. ==== Refs References 1 Egan T.M. From 6 years to 5 days for organ allocation policy change J Heart Lung Transplant 37 2018 675 677 29358011 2 Haywood N. Mehaffey J.H. Kilbourne S. Mannem H. Weder M. Lau C. Influence of broader geographic allograft sharing on outcomes and cost in smaller lung transplant centers J Thorac Cardiovasc Surg 163 2022 339 345 33008575 3 Valapour M. Lehr C.J. Skeans M.A. Smith J.M. Uccellini K. Goff R. OPTN/SRTR 2018 annual data report: lung Am J Transplant 20 Suppl s1 2020 427 508 31898416 4 Grimm J.C. Valero V. III Kilic A. Magruder J.T. Merlo C.A. Shah P.D. Association between prolonged graft ischemia and primary graft failure or survival following lung transplantation JAMA Surg 150 2015 547 553 25874575 5 Loor G. Warnecke G. Villavicencio M.A. Smith M.A. Kukreja J. Ardehali A. Portable normothermic ex-vivo lung perfusion, ventilation, and functional assessment with the Organ Care System on donor lung use for transplantation from extended-criteria donors (EXPAND): a single-arm, pivotal trial Lancet Respir Med 7 2019 975 984 31378427 6 Yeung J.C. Krueger T. Yasufuku K. de Perrot M. Pierre A.F. Waddell T.K. Outcomes after transplantation of lungs preserved for more than 12 h: a retrospective study Lancet Respir Med 5 2017 119 124 27866861

33277026

PMC9754073

NO-CC CODE

2022-12-16 23:26:26

no

J Thorac Cardiovasc Surg. 2022 Jan 31; 163(1):346-347

utf-8

J Thorac Cardiovasc Surg

10.1016/j.jtcvs.2020.10.072

oa_other

==== Front Technol Forecast Soc Change Technol Forecast Soc Change Technological Forecasting and Social Change 0040-1625 0040-1625 The Author(s). Published by Elsevier Inc. S0040-1625(21)00846-5 10.1016/j.techfore.2021.121415 121415 Article Has Covid-19 accelerated opportunities for digital entrepreneurship? An Indian perspective Modgil Sachin a Dwivedi Yogesh K. b Rana Nripendra P. c⁎ Gupta Shivam d Kamble Sachin e a Department of Operations Management, International Management Institute Kolkata, 2/4 C, Judges Court Road, Alipore, Kolkata 700027, West Bengal, India b School of Management, Swansea University, Bay Campus, Swansea SA1 8EN, United Kingdom c College of Business and Economics, Qatar University, P.O. Box 2713 Doha, Qatar d Department of Information Systems, Supply Chain Management & Decision Support, NEOMA Business School, 59 Rue Pierre Taittinger, 51100 Reims, France e EDHEC Business School, 24 Avenue Gustave Delory, 59057 Roubaix, France ⁎ Corresponding author. 3 12 2021 2 2022 3 12 2021 175 121415121415 4 9 2021 27 11 2021 30 11 2021 © 2021 The Author(s) 2021 Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. Covid-19 has challenged many businesses to orient themselves towards digital solutions for their survival. Due to the rising digital wave during Covid-19, there has been a plethora of opportunities for aspiring entrepreneurs to enter the market. Hence, this study focuses on understanding emerging areas and technologies for digital entrepreneurship. This study adopted a qualitative approach with semi-structured interviews through the lens of the diffusion of innovations theory. A total of 23 entrepreneurs responded and presented their views on Covid-19-induced opportunities for digital entrepreneurship. A structured process of open, axial, and selective coding was adopted for the thematic analysis. The study presents a framework based on four promising propositions. Results of the thematic analysis indicate the emergence of digital entrepreneurship opportunities in technology (EdTech, FinTech, cybersecurity), healthcare (diagnostics, virtual care, fitness), entertainment (over the top, gaming, social media), and e-commerce (contactless delivery, payment methods, augmented reality). In this study, entrepreneurs presented their views based on their experience with the platform or technology they operated. To this end, the present study offers implications both for scholars and entrepreneurs working in and aspiring to digital entrepreneurship along with future scope of research. Keywords Digital entrepreneurship Diffusion of innovations Emerging Technologies Covid-19 ==== Body pmc1 Introduction The last two decades have witnessed a trend towards diverse technological changes, not only in business, but also in public systems and on the individual level (Brem et al., 2021; Jafari-Sadeghi et al., 2021). In 2020 and 2021, Covid-19 has been like a storm that led to scaling-up of technological changes and fueling digital entrepreneurship in many parts of the world to address different challenges (Iivari et al., 2020; Secundo et al., 2021). Even in established businesses, the ones those invested in digital operations before Covid-19 fared better than those that did not opt for digital transformation (Volberda et al., 2021; Zahra, 2021). In fact, for many companies today, the continuity of their business depends strongly on their digital capabilities (Datta and Nwankpa, 2021). Even governments are encouraging and moving towards digital innovation and the adoption of new technologies to help the environment and develop new ecosystems (Bai et al., 2021). These digital ecosystems embrace the requirements of digital labor and bots, and the Covid-19 pandemic has accelerated this shift towards greater automation (Brem et al., 2021). Businesses are continually seeking to leverage digital tool, platforms and technologies to maintain uninterrupted operations during crises (Volberda et al., 2021). Additional pressures to improve margins and enhance efficiency drove the need for digital technologies (Zahra, 2021). Many companies from manufacturing, service, and public sectors may have limited access and orientation towards digital technology implementation and monitoring, thus opening the door for third parties to manage digital business operations on behalf of such companies and fueling the demand for digital entrepreneurship in most countries worldwide (Song, 2019; Szalavetz, 2020). In Covid-19, multiple restrictions contributed to an economic slowdown in most of the world during 2020–21, whereas digital entrepreneurial activities witnessed a sharp rise (Bacq et al., 2020; Ratten, 2020; Shareef et al., 2021). Complex and uncertain situations gave rise to entrepreneurial orientations and actions especially addressing the physical challenges with digital technologies. The literature indicates that entrepreneurial activities grow during uncertain times and increase the appetite for risk (Muñoz et al., 2020). In the last two decades, a digital entrepreneurship phenomenon fueled by Covid-19 has driven by technological assets ranging from Internet tools to communication and information technologies (Abubakre et al., 2021; Bai et al., 2021; Secundo et al., 2021). Business opportunities such as transfer of assets, services, or digitalization of organization processes can offer scope for digital entrepreneurship (Jafari-Sadeghi et al., 2021; Song, 2019). The digitalization of business operations has contributed to the emergence of multiple platforms that offer value creation and innovation in business activities focusing on self-employed individuals, small businesses, and entrepreneurs (Brem et al., 2021; Szalavetz, 2020). In the recent pandemic, many mobile-based applications have emerged to monitor the spread of Covid-19 in specific geographical areas and now to track vaccination status and identify nearby vaccination centers along with running the business remotely (Rachul et al., 2020; Sharma et al., 2020). For example, recently developed non-government-regulated portals (for example, cowin.gov.in), such as VaccinateMe from HealthifyMe, have developed a slot-finder mechanism in eleven languages in India (Subramanian, 2021). Another initiative called getjab.in helps individuals find out about Co-WIN slots near their postal code. Similarly, there are other digital entrepreneurial initiatives that are directly linked to Covid-19 and mass vaccination such as BasicFloat.com and Under45.in (Bagcchi, 2021; Subramanian, 2021). According to an estimate, by the year 2019, there were approximately 504 million Internet users, 433 million of whom were children above the age of 12 (ET, 2020b). Covid-19 has further influenced the digital engagement of individuals, employees, and businesses to a significant extent, specifically in 2020 and 2021, which otherwise could have taken several years (De et al., 2020; Dwivedi et al., 2020; Iivari et al., 2020; Papadopoulos et al., 2020). A report by Morgan Stanley estimates that Internet users in India will rise to 914 million by 2027; hence there is a huge scope for online market space such as software solutions, applications, and portals addressing the public crisis and helping businesses (ET, 2020a). In the last few years, multiple platforms (e.g. e-commerce, ride sharing, etc.) have become an inseparable part of life for most of us and are considered a significant sector for any developing economy (Johnston, 2021; Kapoor et al., 2021). Digitally-oriented and technologically-driven platforms also play a key role in enhancing employment levels and innovation culture. The scope for digital entrepreneurship lies in the Internet of things, artificial intelligence, big data analytics, and blockchain technology applications to enhance business competitiveness, performance, and productivity (Dwivedi et al., 2021; Sion, 2019: Zahra, 2021). Before Covid-19, the platforms and opportunities were not considered by many entrepreneurs due to the reason such as lack of familiarity, free-flow movement and no pressure of simplifying business activities. The uncertain and complex environment facilitates the entrepreneurship (Sussan and Acs, 2017). This study took a closer look at the outermost layer of digital entrepreneurship that is visible to consumers by focusing on the following research question “Has Covid-19 accelerated opportunities for digital entrepreneurship?” This study contributes to diffusion of innovations theory with regard to digital entrepreneurship and its underlying business expansion opportunities. This study observed the different factors that influence the adoption of an innovation and indicate how innovation is perceived better than the existing programs or products, how consistently innovation stand with the value and experiences of potential adopters, how difficult or easy it is to use, ability of the innovation to an experiment and the extent to which it can offer tangible results. The rest of this paper is organized in five sections. Underlying elements are presented in section two. The research design is presented in section three. The findings of the study and propositions along with the framework are presented in section four. The discussion on the findings is presented in section five. The implications are presented in Section 6, whereas Section 7 concludes the study. 2 Emerging technologies and digital entrepreneurship In this era of digital disruption and internet connectivity, developing economies need to take advantage of emerging technologies to strengthen digital entrepreneurship through innovative solutions to meet societal needs and problems (Wang et al., 2021). The role of emerging technologies and digital entrepreneurship is therefore presented below. 2.1 Emerging technologies Today, companies embrace the adoption and employment of digital tools in their businesses to create and modify existing business processes. Apart from benefiting businesses directly, emerging technologies help organizations to develop their workplace culture and enhance the consumer experience (Kamble et al., 2021). Digital technologies enable companies to reassess their business operations, align resources, and develop capabilities to create a framework to drive innovation in business activities (Schiavone et al., 2021). In addition to business use, emerging technologies have great potential for the general public, and many companies have started to develop services in that direction. For example, CivilCops, a company founded in 2017, exploits big data and artificial intelligence (AI) to fast forward the complaint and resolution system in the public domain (Rana et al., 2016). CivilCops works with the government and can be reached with the swipe of a key. In smart city management, CivilCops utilizes data to provide actionable insights to continually improve smart city operations. It uses machine-learning algorithms to analyze the nature of complaints and map the particular department in order to take the necessary action. Another company, Oxfordcaps, employs technology to improve the student living experience. Oxfordcaps uses AI and machine learning models to enable searches and book accommodation without physical visits and allows hassle-free signatures and payments. Similarly, Rezo.ai makes use of conversations between brands and their customers and automates those conversations to analyze customer concerns, intent, and queries as supported by recent literature (Dwivedi et al., 2021). To do this, Rezo.ai feeds textual dialog into its AI-enabled platform and automates the customer journey with the help of a machine-learning algorithm (Mint, 2019). Another organization, ZunRoof, uses virtual reality, big data, Internet of things, 3D printing, and image processing to generate electricity from solar power and design a solar rooftop system for households needs. Therefore, it can be observed from recently launched digital enterprises such as CivilCops, Oxfordcaps, and ZunRoof, that emerging technologies are not only solving business and public problems, but also contributing to the development of sustainable and innovative ecosystems for the planet, without consuming many resources (Mint, 2019). 2.2 Digital entrepreneurship A strong need towards digital technologies in the last two decades has resulted in many digital artifacts, digital platforms, and digital infrastructure development initiatives by public and private entities. A digital artifact is defined as a digital element, application, or content interrelated to a product or service that facilitates a particular functionality for the benefit of the end user (Liu et al., 2021). The decoupling of information from its physical product has led to the rise of services in digital artifacts (Barrett et al., 2015; Islam et al., 2020). Such applications cover a wide range of products ranging from smartphones, toys, and automobiles to apparel. Therefore, digital artifacts can be classified as software/hardware element on physical products or as a part of an ecosystem that function on a digital platform and offer many opportunities for digital entrepreneurship (Schiavone et al., 2021). A digital platform is defined as shared space to host services and an architecture that provides complementary offerings along with digital artifacts. Digital platforms offer a plethora of opportunities for entrepreneurs to develop complementary products and services. Digital platforms are attractive to entrepreneurs in terms of production, marketing, and distribution of services (Nambisan, 2017; Srinivasan and Venkatraman, 2018). Digital infrastructure is a set of digital tools, technology, and systems (big data, 3D printing, online communities, etc.) that offers computing capabilities, collaboration, and communication capabilities for innovative solutions to organizational problems (Elia et al., 2020). Digital infrastructure helps entrepreneurs follow the democratic process of opportunity with consideration given to concept testing, funding, and launch (Sussan and Acs, 2017). Digital entrepreneurship has seen a sincere persuasion in recent times due to the availability of technologies such as cloud computing, big data analytics, and market spaces. 2.3 Diffusion of innovation For entrepreneurs, it is very critical to perform rigorous market research to understand and design products or services those are innovative and unique in nature. In the age of Internet and connectivity, individuals are excited about new and innovative concepts that can solve business problems (Zajicek and Meyers, 2018). Hence, the diffusion of innovation (DoI) theory is most suitable to explore the possibilities of digital entrepreneurship. The DoI helps entrepreneurs to visualize how, why and at what rate novel concepts and technology extents (Rogers et al., 2014; Rogers, 1995, 1962). DoI helps entrepreneurs to analyze and predict the consumers’ adoption behavior about their service or product (Marcati et al., 2008). The development and adoption of new idea or service takes time and adoption by earlier customers represent different characteristics than who adopts innovation later (Cao and Shi, 2021). Hence, the digital entrepreneurs need to have appropriate understanding of every element that can facilitate or hinder the adoption of innovation (Abubakre, 2021). The adopters according to DoI can be classified into innovators (first mover), early adopters (those embrace the change and new ideas), early majority (embrace and adopt the innovative ideas before it research to mass and a evidence is needed that innovation works before entrepreneurs believe its worth), late majority (represented by those, who are uncertain about the idea and change and adopt the idea after being generally accepted by the population) and laggards (conservative and traditional entrepreneurs who are the last to shift to new technologies) (Rogers, 1962). Therefore, digital entrepreneurship presents a scope not only for unique and fist hand ideas in the market, but also to the matured markets such as website and content development enterprises. Furthermore, DoI can make unique contribution in visualizing the adoption and expectation of consumers to design and develop innovative propositions. 3 Research methodology To examine the research questions regarding digital entrepreneurship presented in this study, a qualitative approach is adopted. Since, there is lack of established scale with reference to Covid-19 or very complex environment accelerated opportunities for digital entrepreneurship. Therefore, to address the research questions, it was suitable to start with qualitative study. Due to the busy work schedule of entrepreneurs, limited and well-designed semi-structured interviews were conducted. The qualitative study is best suited to understanding the views of working professionals to obtain better insights. Therefore, this study presents insights for digital entrepreneurs operating in complex and uncertain environments such as Covid-19. 3.1 Data collection Entrepreneurs from different organizations and sectors servicing local and regional markets were interviewed. The semi-structured interview schedule was developed based on the guidelines of Leech (2002) and McCracken (1988). In total eight questions were developed (see Appendix A), focusing on digital entrepreneurship and technological acceleration with Covid-19 as the impetus for innovation and unique services to solve public and private sector problems. In total, 151 entrepreneurs in the digital field were contacted through LinkedIn in October and November 2020. First, the concept of the study was introduced and requested a time and date for a 30 – 45 min interview. After three consecutive follow-ups in December 2020 and January 2021, 29 respondents were interviewed. Careful transcription and filtering finally resulted in 23 responses. It has been also observed that almost after 23 responses, there was saturation of responses. Therefore, overall 23 responses were finalized for further analysis. Table 1 indicates the profile of the interviewed entrepreneurs. These respondents were further mapped as R1 to R23 to maintain anonymity. Data were collected from India that presents potential for digital entrepreneurship due to an increasing smartphone-user base. The current smartphone-user population is approximately 50% in India. This study therefore considered the scope of digital entrepreneurship for the emerging needs of local and global businesses.Table 1 Detail of respondents. Table 1Respondent profile no. Respondent code Job title Domain of work Years of experience Year of establishment 1 R1 Managing Director Real Estate 8 2015 2 R2 Founder & CEO Trading 9 2014 3 R3 Co-Founder E-Bills 6 2017 4 R4 Founder Clothing 8 2016 5 R5 Co-founder E-Fitness 7 2017 6 R6 Director Entertainment 7 2018 7 R7 Chairman Education 9 2015 8 R8 Co-Director Website Development 9 2012 9 R9 Director Gaming 7 2014 10 R10 CEO FinTech 6 2015 11 R11 Co-Founder E-Health 8 2015 12 R12 Founder E-Commerce 6 2016 13 R13 President Entertainment 8 2013 14 R14 Co-Director Payments 7 2015 15 R15 Founder Digital Marketing 8 2017 16 R16 CEO Education 7 2018 17 R17 CIO App Development 5 2017 18 R18 Co-founder Digital Advertising 8 2014 19 R19 Founder EdTech 6 2016 20 R20 Co-founder E-Commerce 7 2017 21 R21 Director E-Health Services 6 2018 22 R22 CTO Technical Support 5 2017 23 R23 Founder Cyber Security 7 2015 3.2 Data analysis After conducting the interviews, a verbatim transcription and thematic coding mechanism was applied to extract themes and sub-themes. Fig. 1 indicates the research design. The manuscript utilized transcription a number of times to ensure internal consistency. As a measure of triangulation, the study also compared the themes and sub-themes through available secondary data.Fig. 1 Research design stages. Fig. 1 A three-layered coding mechanism was followed to analyze the raw data. First, this study extracts open codes from the interview responses. Second, open codes were further mapped to the emerging axial codes. Finally, axial codes were mapped to the selective coding. After viewing and funneling down to selective codes, a triangulation approach was applied to verify and validate the themes that emerged from the data given in Table 2 .Table 2 Triangulation approach. Table 2Industry reports Research articles Data gathered Mapping to themes UNCTAD (2021)-Despite slowing economic activities, digital transformation accelerated in e-commerce with a rise in global trade from 14% in 2019 to 17% in 2020. New trends such as monitoring of social distancing and augmented realty have offered opportunities for entrepreneurs. Johnston (2021)-Electronic world trade platforms help unlock trade in developing countries and offer many opportunities for digital entrepreneurs. R20-There was gap of touch and feel that was even wider due to the fact that shops and malls were shut for a long time. Hence, technologies such as augmented reality have appeared as saviors for an industry such as e-commerce. Contactless delivery, Augmented reality McKinsey and Company (2021)-In digital health services, one needs to have experts from the technology field as well as healthcare and regulatory officers to deliver the diagnosis or virtual care service. Zajicek & Meyers (2018) - Digital healthcare requires innovative solutions and creates value defined by the users in terms of service, platform, and models. Digital home care is on the rise due to patient comfort. R11-As most of the population is health cautious due to the fear of Covid-19, the scope is huge, and people are embracing digital solutions in the healthcare and fitness domains. E-fitness Virtual care PwC (2020)—Covid-19 has amplified and shifted consumer behavior due to the high demand for digitalization and due to social distancing and mobility restrictions. Madnani et al. (2020) - The world has become remote, virtual, and more streamlined. Therefore, consumers have developed customized entertainment consumption behaviors. R6-As people stay at home during 2020 and 2021, over-the-top platform revenue surged by 26% in 2020 and is expected to double in 2024. Over the top KPMG (2020)-EdTech technology is transforming traditional education from paper and pencil to digital and video-assisted learning. Iivari et al. (2020)-Covid-19 has created a variety of digital divides through the facility of EdTech, which can better develop skills and competencies that may be helpful in career options later on. R16-The impact of Covid-19 has been felt most in the education field. This changed the fate of the EdTech sector overnight with the indefinite closure of educational institutions. Educational technology 4 Findings 4.1 Technology 4.1.1 Educational technology Digital adoption in school and college educational institutions has accelerated greatly due to Covid-19. Covid-19 forced most education systems worldwide to go remote and opt for virtual learning overnight. This has fueled digital entrepreneurship in the majority of educational fields (Iivari et al., 2020). Many startups have been funded by venture capitalists due to their potential as well as trust in the performance of these platforms even in the post-pandemic scenario. In the words of R7, “apart from helping the learner, educational technology (EdTech) widens the scope for customized learning for the scholar on the basis of his or her skills, interests, and strengths. EdTech is offering a platform to empower educators with high-tech tools to innovate in terms of student-learning styles and make the teaching and learning process more effective”. Some of the start-ups in the EdTech field focus on certain areas such as childhood development, where they regionalize the curriculum and stories centered around the local culture. A few digital entrepreneurs in the education field have also brought in innovations such as gaming. These platforms are personalizing the learning experience with the aim of problem solving and critical thinking. EdTech entrepreneurs are expanding their scope to coding, robotics, classical dance, and musical instruments to make it more diversified. 4.1.2 Financial technology The financial technology (FinTech) sector has continued to expand its services during Covid-19, especially in emerging markets. Access to financial services by corporations and individuals can help an economy grow and increase income levels along with improving resilience and quality of life. FinTech platforms are facilitating services to reduce their cost of operations and to reach out to as many people as possible. More importantly, FinTech platforms are reducing face-to-face interactions and keeping pace with the economy (Vasenska et al., 2021). In the words of R10, “the online payment industry has witnessed a multi-fold jump in its subscriptions, which demonstrates the potential for payment and banking technology start-ups. Digital technologies have a huge scope in the fields of payment collections, quick loan disbursement, and vendor payouts among other concepts”. The industry as a whole offers space for innovation and critical insights for policymakers and regulators, while facilitating financial stability, management of customers and investors, etc. FinTech also contributes to relief efforts in a pandemic-like situation and offers multiple services to micro, small and medium-sized (MSMEs) businesses. Even though Covid-19 has accelerated digital initiatives across sectors with the increasing flow of new ventures, some start-ups are still struggling with their financial positions and must be mindful of their actions. 4.1.3 Cybersecurity Businesses around the world have been focusing on digitization and the large-scale migration towards the cloud has further led to a significant need for secure networks. With the current pandemic outbreak, most companies have moved to a remote working style and have opened the door for many start-ups in the cybersecurity field (Lallie et al., 2021). According to R23, “due to the work-from-home (WFH) environment during Covid-19, many executives from IT, operations, and security teams are facing cybersecurity challenges and are concerned about how they can prevent cyberattacks and the propagation of malware. Therefore, the challenge for most start-ups is to provide innovative, customized, simple, and affordable cybersecurity solutions for even the smallest organization”. On one hand, firms are keeping pace with the competition to offer better products and services to their customers, while on the other hand there are concerns regarding threats in cyberspace. The importance of cybersecurity has been underscored by different incidents in various industries. Many start-ups have emerged, thanks to the application of AI and machine learning-based models that can evaluate risk using different methods. Table 3 describes the interview data and open, axial, and selective codes for technology.Table 3 Thematic coding for ‘technology’. Table 3Sample quotes from interviews and open code Profile of respondent Axial code Selective code “The digital revolution has impacted most of us, and K-12 education is no exception. The digitization of education on one hand provides opportunities for start-ups to offer innovative and creative solutions, while on the other hand it fosters self-learning, collaboration, and creativity” (Open code: any-time learning from digital education). R7 Educational technology Technology “Compared to the traditional teaching and learning system, especially competitive exams for public service and entry to higher education, learners have more flexibility when using the digital method due to its extensive coverage of topics and its easy updating in digital form from a start-up point of view” (Open code: variety of services for learners and easy to update). R1 “Today, in the era of applications and smartphones, we have developed an AI-based decision engine that evaluates the loan application in minutes and up to Rs. 5 Lakh can be transferred to the respective account. To ease any difficulty for the customers, we have enabled the repayment options through platforms such as PhonePe” (Open code: AI-based platform to analyze and process loan applications). R3 Financial technology “Due to the high volume of transactions in businesses such as insurance, e-commerce, and banks, the reconciliation process has been automated. Being a third party, we ensure that every transaction is accurately accounted for with minimal settlement time by deploying AI models to connect the payment gateways, banks, and vendor order system which further helps identify any discrepancies” (Open code: a high volume of business requires the assistance of financial technologies). R11 “Around 50% of all the security attacks concerned small and medium enterprises, which may affect their business significantly. Hackers are capitalizing on Covid-19-like situations for geopolitical supremacy, financial gain and for reputation reasons” (Open code: for different reasons, hackers are planning cyber-attacks). R19 Cybersecurity “Industries ranging from healthcare to defense, education, hospitality, and banking are prone to cyber threats. Hence, secure networks are the need of the hour for every business. Not only should the business have on-premises threat detection and mitigation mechanism, but it should also be careful about cybersecurity capabilities across the cloud” (Open code: the data of a business require multi-layer cyber security mechanisms). R16 4.2 Healthcare 4.2.1 Diagnostics Covid-19 impacted the healthcare sector the most, thus offering opportunities for digital entrepreneurship in this sector. From medicine and vaccine manufacturers to labs, pharmacies, and hospitals, each stakeholder is compelled to innovate (Jnr, 2020). Most healthcare operators have to be quick during Covid-19 to stay relevant. Sometime there is a large gap between the actual incidence of Covid-19 cases and the reporting, so technology-driven solutions such as AI and machine learning can speed up the process of clinical diagnosis and offer opportunities for digital entrepreneurship. In the words of R21, “clinicians today are facing the challenge of a complex scale of disease and large available information outlets, so making accurate predictions about a particular diagnosis is critical. Digital start-ups today are offering a wide array of solutions from testing to imaging to pathology to offer personalized health insights”. Many start-ups are utilizing AI-based medical imaging platforms for diagnosis. 4.2.2 Virtual care Virtual care focuses on services in terms of interactions with healthcare professionals and online consulting with clinicians. This sector offers significant potential to address the issues of healthcare accessibility. Improved management of a disease requires monitoring of the acute condition and daily routine. This is used to design better care for patients in their homes. During Covid-19, the government placed great emphasis on telemedicine and witnessed a significant reduction in consultation time and greater efficacy through different digital platforms (Janssen et al., 2018). Telemedicine platforms are quite popular where well-qualified practicing clinicians are used and patients also have the option of getting a consultation from a selected clinician (Schiavone et al., 2021). Virtual care platforms are further integrated with e-pharmacies, where in one click patients and customers can order medicines after a prescription is verified. R5 stated that “digital entrepreneurship in virtual care is converting one-time patients into regular consumers thanks to benefits such as ‘an individual can get treatment from home’, ‘access to special physicians’, ‘reduced cost of treatment’, and ‘flexible scheduling and appointments’. These platforms also make the job of clinicians easy and offer room for innovative approaches in virtual care delivery”. Virtual products and care platforms help reduce the wait time for the individual having the consultation online (Algharabat et al., 2017). Additionally, in view of the Covid-19 outbreak, people are avoiding visiting hospitals due to the fear of contracting another illness. 4.2.3 Wellness Developing immunity and staying fit have been buzz words throughout the Covid-19 crisis. Many people have developed healthy habits and are concerned about their fitness. Many people have searched how to stay fit and develop strong immunity on the Internet. In this regard, fitness technology has witnessed exponential growth and many fitness clubs and gyms are conducting live sessions. Since most gyms were closed due to the lockdown and surge in Covid-19 cases, this technology has opened a plethora of opportunities for digital fitness aggregators, trainers, and fitness aspirants (Bentlage et al., 2020). Digital yoga platforms have also emerged, and many start-ups have observed significant growth compared to earlier years. According to R11, “the technology-driven fitness, yoga, and diet field has created its market and continues to do so due to the increased awareness of people becoming pre-emptive about health and measures to improve from time to time”. Recent events have resulted in pay cuts, anxiety, and a significant decrease in movement, and people are looking for easy, simple, affordable, and accessible wellness solutions for their health. For digital start-ups, online advertisements for fitness and wellness were low cost due to the shutdown of many marketplaces, and it was therefore easier to do advertising. This resulted in a low cost of acquisition per person. Table 4 presents the interview data and open, axial, and selective code for healthcare.Table 4 Thematic coding for ‘healthcare’. Table 4Sample quotes from the interviews and open code Profile of respondent Axial code Selective code “Awaking to the impact of Covid-19, many emerging economies made commitments to enhance healthcare infrastructure to tackle both Covid and non-Covid health issues. This has given birth to numerous digital health start-ups and operational fundraising” (Open code: digital healthcare services for both Covid and non-Covid related health issues). R11 Diagnostic Healthcare “Cardiovascular issues are the leading causes of death. Therefore, timely diagnosis and treatment is important. With the help of image analytics and MRI-integrated AI, algorithms help assess the arterial function of a patient. With the help of a web-based platform, one can detect and diagnose cardiovascular disease in the early stage from the comfort of the patient's home” (Open code: web-based platforms that can facilitate online diagnostics). R5 “In the last two years, we have witnessed a significant increase in telehealth users, where both image and video-based service are available for both telemedicine and remote patient monitoring. Telehealth is more efficient in terms of cost and makes it convenient for most patients due to their comfort at home” (Open code: due to staying at home, people are opting for telemedicine and virtual treatment via digital platforms). R21 Virtual care “In virtual care, companies are exploring and providing patient-to-clinician, clinician-to-patient and multi-party video visits, since healthcare may require different specializations to come together for a particular disease” (Open code: facilitating patient, clinician, and multiparty communication through web-based video and images). R8 “We have developed a platform that uses a pedometer integrated with gamification (number of coins earned, number of levels achieved, and benchmarking through a leaderboard) and gratification (through prizes and rewards) to encourage people to move more and have a healthier lifestyle” (Open code: motivating individuals through gamification to stay fit). R12 Wellness “The problem is not how to keep aspirants moving and exercising routinely, but rather how to motivate them to be fit. Being physically distant for quite some time, people now realize the importance of community. Hence coaching and fitness sessions need to be designed to make things more intimate through the digital route” (Open code: design of an interactive and intimate digital platform). R10 4.3 Entertainment 4.3.1 Over the top Lockdown and other measures to curb Covid-19 have brought after-effects to the media and entertainment industry, specifically for film-making, theme parks, and entertainment. However, this has opened up the opportunity for greater digital media consumption. During the shutdown and lockdown, most parts of the world witnessed the growth of over the top (OTT) television and digital platforms due to the ban on outdoor activities and social distancing norms (Madnani et al., 2020). R6 pointed out that, “Digital media consumption, specifically OTT, has witnessed a surge in subscription through Netflix, Amazon Prime, Zee5, etc. both in terms of time spent and new consumers. This could result in a shift from mobile to television screens with the ease of internet connections”. Connected via an OTT platform, the advertisement video-on-demand, subscription video-on-demand, and freemium models are gaining traction and have doubled in the past years. Innovation is the hygiene factor for OTT players and digital entrepreneurs. Hence, organizations in digital businesses such as OTT have to evaluate performance based on multiple facets and set-up in-house laboratories to stay innovative and customer focused. 4.3.2 Gaming Historically, games have been an integral part of human culture, and over time they have changed from being physical to virtual. Exponential industry growth has challenged traditional forms of entertainment and games. The online gaming industry includes stakeholders such as game developers, software developers, hardware developers, game publishers, distributors, and retailers (Amin et al., 2020). Consumers are mostly aware of the distributors, such as Google Play and Game Stop, but are unaware of game publishers and developers, where most entrepreneurship opportunities exist. Out of the three mobile, computer, and console-based gaming platforms, mobile gaming has the largest share with around 45% and receives 80% of revenues. In the opinion of R9, “interestingly, mobile gaming applications have become the most popular applications after social media and shopping applications. The success of mobile-based online gaming can be attributed to new loyalty programs that keep players coming back to the platforms. Players can earn rewards in the form of points and other forms of staying in the game, such as a subscription to watch free advertisements”. To make it faster and easily accessible to consumers, companies are moving to the cloud, and decent internet connectivity ensures the streaming of games rather than playing via a console. In future, the game will be able to generate auto-content and be customized according to the player's personality and gaming style. 4.3.3 Social media Today, social media has surpassed the phase of connecting family and friends, and now helps businesses connect with their customers. This offers the opportunity for third parties to operate on behalf of focal organization to design, operate, and promote their products on social media. New age entrepreneurs are using the internet of things to promote products and automate workflows to engage customers. Other emerging technologies such as data storage via cloud computing, networks, and software management are covered by third party entrepreneurs to save costs and increase productivity (Song, 2019; Szalavetz, 2020). In the words of R15, “the close connection through social media helps companies monitor consumer behavior and design platform-centered insights to drive business. Advertising and marketing solutions other than social media messages address the key concerns of many organizations”. Organizations are utilizing social media to map their competitors and accumulate knowledge that can be helpful in designing and offering new products and services in the future. Entrepreneurs that assist product and service companies are utilizing innovative and interactive techniques to drive traffic on their pages. Table 5 describes the interview data and open, axial, and selective code for entertainment.Table 5 Thematic coding for ‘entertainment’. Table 5Sample quotes from the interviews and open code Profile of respondent Axial code Selective code “The growing consumption of digital media is challenging most entrepreneurs to experiment with new models of content consumption and hence spend on advertising as well as to shift from the traditional television system towards digital media platforms. Additionally, as consumption is mostly through smartphones, mobile advertising is expected to grow exponentially in the future on OTT platforms” (Open code: shift from traditional to digital advertising). R15 Over the top Entertainment “The OTT landscape offers digital entrepreneurship aggregation in the field of content creation as compared to digital platforms such as YouTube, Voot, Netflix, etc.” (Open Code: the scope for digital entrepreneurship in the domain of content creation). R9 “With increased smartphone usage, there is rising demand for different types of games that are interoperable on Android, iOS, and Microsoft Windows. Single and multi-player games in both mobile and computer formats are in high demand due to the paradigm shift from console device- oriented gaming” (Open code: shift from console to mobile-oriented gaming). R20 Gaming “Due to increasing screen time on mobile phones and computers, it has been observed that young people play games alone and with peers and frequently look for updated versions or new games. To address this demand from consumers, companies can adopt a platform as a service, where the creation of games will be easier in a short span of time” (Open code: platform as a service facilitates easy and frequent updating of games). R12 “Due to improved connectivity and internet penetration, now most start-ups and established organizations are present on social media, and they believe it is one of the easiest ways to connect to their customers, leading to the fact that content is king and can influence customers” (Open code: connecting to customers through social media). R14 Social media “To be seen by consumers on social media, one should have good content, and it will trend. Today, many established organizations are asking for help from digital entrepreneurs and agencies in the form of posts, blogs, celebrities, and stories” (Open code: social media offers opportunities for creating stories, blogs, and posts). R23 4.4 E-commerce 4.4.1 Contactless delivery Covid-19 impacted the lives of humans to the extent that now people fear going out even to buy items that are essential to them. Therefore, instead of visiting local stores and fashion outlets, people prefer online platforms even for their daily needs. Apart from convenience, these platforms offer safety in terms of contactless delivery and payment. With the rise in e-commerce in the last decade fueled by Covid-19, there has been a change in the style of home deliveries and doing business. Contactless deliveries have become the new norm and retailers who do not offer contactless deliveries are not part of the game anymore (Johnston, 2021). The contactless delivery process avoids interaction between employees and customers. The employee confirms the delivery of the order with a picture shared with the customer via a mobile application. The customer is reminded through alerts and messages to pick up the order outside the door at his or her convenience. Similarly, R12 pointed out that, “Contactless deliveries are helpful for both customers and organizations for safety reasons. Additionally, to build confidence among customers, some companies are also displaying the health status of delivery boys and related store staff through the order tracking screen. The scope for digital entrepreneurship lies in contactless pickups from retail, fulfillment centers, and dark stores. From label generation to order collection and order assignment, everything needs to be done without contact and with minimum face-to-face interaction”. Any lack of or compromise in safety at a pick-up point or at other stages in the supply chain can lead to a chain reaction and an increase in the level of risk. Different models are prevalent for contactless delivery in e-commerce, such as one-time passwords, photo captures, or online links. 4.4.2 Payment methods Setting up a business online is a current trend due to increased smartphone penetration throughout the world in the past decade. However, fund flows are equally important to any business for its other activities. The smooth flow of funds in an e-commerce environment can be realized using payment platforms/methods. Gone are the days when organizations used to accept cash payments only. With the advent of technology, a wide range of UPIs, e-wallets, and mobile payments are now available to suit the needs of diverse customers (Vasenska et al., 2021). In addition, R20 stated that “nowadays, businesses not only in e-commerce but also in the physical store format are using more than one payment method for the convenience of customers, and this helps with sales conversions. Multiple payment methods are also helpful in the scenario where one server is down or sometimes does not work, and a second payment gateway can be helpful in the purchase process”. Many payment aggregators have emerged with the growth of online shopping and e-commerce. The role of payment gateways/aggregators is to support and link the transaction between the customer, bank, and seller. 4.4.3 Augmented reality In the e-commerce environment, customers are continuously shopping, and sometimes customers find that a product is not suitable to their requirements and return it. The top reasons for returning consumer goods are wrong size, fit, or color. Technology such as augmented reality (AR) can be very helpful in addressing these challenges due to its ability to expand the physical world by adding a layer of digital transformation (Datta and Nwankpa, 2021). Employing AR offers a view of the physical/real environment facilitated by superimposed computer-generated images to better understand product shape, size, and fit. In the words of R4, “AR helps e-commerce consumers know, understand, and make up their mind about what they are ordering and how accurately products ranging from cosmetics to clothing will suit their bodies. Virtual try-on solutions help customers feel and look before adding the item to their shopping cart. Today, brands ranging from jewelry to eyeglasses are employing AR for customer satisfaction and bringing in more traffic”. Furthermore, AR is applied to e-commerce through interactive user guides to make customers understand product flows and how they work. Many AR-based users’ manual applications present the product in real life through animation. Table 6 describes the interview data and open, axial, and selective code for e-commerce.Table 6 Thematic coding for ‘e-commerce’. Table 6Sample quotes from the interviews and open code Profile of respondent Axial code Selective code “Today, if your business is not online and not connected with the smartphone of the user, you may be out of business soon. When designing logistics management for a particular supply chain, contactless delivery monitoring is also very much a part of it and creates opportunities for digital start-ups” (Open code: connecting businesses to mobile phones to track contactless delivery). R2 Contactless delivery E-commerce “Different companies operate on different models for contactless delivery. For example, a few companies share notifications with customers that their order is out for delivery and they can view the estimated time of arrival. Other models work with a one-time password (OTP) shared with the customer when the shipment is out for delivery, and this OTP is confirmed by the delivery executive” (Open code: different models for contactless delivery). R14 “With the rise in e-commerce, companies are increasingly exploiting online payments. E-wallets and mobile payments are the most common ones compared to cash on delivery. Since e-commerce has become more phone-streamed, customers are finding mobile payment solutions convenient” (Open code: importance of multiple payment methods). R15 Payment methods “For processing the payment in e-commerce, the payment gateway is an essential requirement which further needs to be integrated with the website and account of an organization. This payment gateway facilitates the link between customer and bank and notifies the seller of the amount credited for a particular transaction” (Open code: the payment gateway role as aggregator). R17 “For the product-driven industry, augmented reality (AR) can be very helpful. In e-commerce, AR allows customers to see products at their convenience as well as their suitability for the given purpose before an actual purchase. AR can help customers choose the right product the first time and avoid multiple returns as witnessed with normal online purchases” (Open code: giving more confidence to consumers before an actual purchase). R6 Augmented reality “In our industry it is observed that many products are returned due to its mismatch to expectations of the customer. It also believed that e-commerce will be rising. For example, a customer might purchase a cupboard and later find that it doesn't fit in the space they have, therefore a gap exists, where in technological innovation can be helpful both for customer, organization and sustainability point of view” (Open code: choosing the right product on e-commerce platform via AR). R16 5 Discussion The study followed a semi-structured interview approach to understand diverse areas of digital entrepreneurship opportunities, and this revealed some stimulating implications for extending DoI theory in trying times such as during the Covid-19 pandemic. This study assimilates the theory and literature on innovation, entrepreneurship, and technology. Earlier scholars have emphasized entrepreneurship opportunities in the physical mode of business due to wider availability in the physical world and similar customer preference for physical products. With the changes fueled by Covid-19, there has been a significant increase in domains for digital entrepreneurship with a consistent focus on innovation (Brem et al., 2021; Volberda et al., 2021). Therefore, the study theorize diffusion of innovations theory through the lens of digital entrepreneurship in the uncertain, complex, and concerned environment created by Covid-19. 5.1 Spotting trends in covid-19-induced digital entrepreneurship The complex and uncertain Covid-19 era pushed organizations to a tipping point in terms of technology adoption for transforming their businesses. This digital transformation process of businesses has opened the door for many start-ups and digital entrepreneurs to seize the opportunities (Datta and Nwankpa, 2021). Customers along with companies have shifted online significantly, be the students and educational institutions or a consumer banking segment, and towards other changing virtual workspace requirements such as working from home. With online education in terms of regular classes and competitive examinations, many new start-ups have emerged to cater to different needs of learners and aspirants. Running businesses and households has given rise to financial technologies. This change from a physical to a virtual workplace has increased the pressure of cybersecurity on organizations and has been impacted by hacking incidents in 2020 and 2021 for multiple reasons. Hence, this study proposes: P1: The technological changes induced by the Covid-19 environment offer significant scope for digital entrepreneurship in the educational, financial, and cybersecurity fields. Health has been the most sensitive issue for most of the world during the Covid-19 period. To go beyond the traditional healthcare system, the trends in diagnosis and examination, healthcare technologies and digital entrepreneurship have changed the landscape forever. Due to increased smartphone penetration in most locations, digital healthcare enterprises have been able to find customers easily. Today, enterprises in healthcare sectors are using digital technologies beyond telemedicine, such as laboratory tests, diagnostics, virtual care, and fitness (Kim et al., 2016). A series of digital enterprises have come up with fitness and yoga programs that engage their consumers digitally with uniquely and innovatively-designed programs. Hence, this study proposes: P2: Digital healthcare services are in demand due to scheduling convenience, the comfort of home, avoidance of crowds at physical facilities and the low chance of error due to the use of technologies such as big data analytics, artificial intelligence, and virtual reality. Many individuals experienced quarantine during Covid-19 due to traveling, for their own safety as well as government mandate or because of contracting Covid-19. Out of fear, people have stayed home for longer periods. These intricacies led to the consumption of more television and spending more time on social media. As movie theaters were closed, people were not able to watch movies, so over-the-top (OTT) platforms were accepted quickly by audiences to meet their entertainment needs (Madnani et al., 2020). These platforms offer opportunities in the field of digital advertising. Young people fond of sports and games shifted to online games, which opened up opportunities for entrepreneurs in this field to design and develop games to meet the increasing demand. Gaming has changed a lot over the last two decades from being console-oriented to computer-controlled and now to mobile-based, where it can be in a single or multi-player format. The gaming sector has witnessed a boom in multiple categories in the recent past. Hence, this study proposes: P3: Entertainment as an essential part of human recreation is evolving and presenting entrepreneurs with opportunities in the field of social media engagement, next-generation gaming, and innovative ways to present video content. Lockdowns have become the new normal due to multiple waves of deadly virus, and businesses and consumers have decided to go digital for their survival. This has led to the rise of e-commerce for purchasing goods and services (Algharabat et al., 2017; Carter et al., 2016). On the one hand, many e-commerce-based platforms are struggling to respond to the sudden surge in demand, while on the other hand they are facing a huge issue of product returns due to a mismatch with customer expectations such as the wrong size or color. This has led many companies to adopt augmented reality to minimize the product return rate and enhance customer satisfaction. On one hand, people are shopping for essentials as well as luxury products and want delivery to their home, while on the other hand they do not want to come in contact with anyone from the outside due to their fear of contracting Covid-19. This has resulted in the birth of contactless delivery (Johnston, 2021). This process of online shopping is not complete until payment is made on the platform, so due to the diverse customer base and for payment comfort, companies have come up with different payment methods. Therefore, this study proposes: P4: Next generation e-commerce has opened up opportunities for digital enterprises in the fields of improving and monitoring contactless delivery, ensuring multiple options for successful payment on the consumer end, and reducing the return rate by employing augmented reality for a better look, fit, and feel according to the customer. The Fig. 2 below illustrates the emerging fields for digital entrepreneurship propagated through Covid-19.Fig. 2 Emerging digital entrepreneurship fields in the era of Covid-19. Fig. 2 6 Implications This section presents the implications for theory and practice along with limitations and scope for future research. 6.1 Implications for theory This study contributes to the literature in three ways. First, it mines the information from seasoned entrepreneurs to extract the sectors and areas of significant potential for digital business in a structured manner through the lens of DoI. This theory helps entrepreneurs visualize how, why and at what rate novel concepts and technology are progressing (Rogers et al., 2014; Rogers, 1995, 1962) and can facilitate in taking certain business decisions. DoI helps entrepreneurs analyze and predict customer adoption behavior influenced by a complex and long event such as Covid-19 with regard to their service or product (Marcati et al., 2008) that can actually help them to improve upon the products and services offered. The development and adoption of new ideas or services takes time, and early customer adoption has different characteristics than later ones (Cao and Shi, 2021). Therefore, digital entrepreneurs need to have an appropriate understanding of each element that can facilitate or hinder the adoption of innovation (Abubakre, 2021). By employing the services of digital entrepreneurs, larger companies can further influence international business practices and culture. Digital entrepreneurship can be viewed as a capability or facilitator that can transfer the physical value chain to the virtual realm and cater to consumer needs during crises and complex times more effectively. Entrepreneurs from different fields conceptualized emerging opportunities for digital enterprises. Covid-19-induced digital entrepreneurship is facilitating and infusing innovation capabilities into the traditional style of production, marketing, and delivery along with new channels of communication with critical stakeholders. Covid-19 has also fueled digital start-ups, due to the increasing hygiene demands through digital solutions such as telemedicine and virtual care (Sussan and Acs, 2017). This hygiene requirement will be expected in the post-Covid era as well, so there is room for increase in digital entrepreneurship even post-Covid world. Second, on the basis of data collected, the study established four propositions that need to be further tested via a grounded approach. Theories such as resource-based view, task-technology fit can be considered to further test the claims. Third, this study proposed a framework indicating different areas ranging from technologies, healthcare, entertainment, and e-commerce that are leading in the race for opportunities for digital entrepreneurs. Additionally, the study highlights the role of diffusion of innovations theory, specifically in relation to digital entrepreneurship and digital start-ups, for their ultimate success and monitoring the business environment continuously. This study presents emerging areas in a particular domain that can be further explored by aspiring digital entrepreneurs in the digital realm. It offers a list of detailed areas in a particular domain with key insights, where innovation can play a key role in the next few years and benefit budding entrepreneurs in the digital field (Brem et al., 2021; Kim et al., 2016). Technology in the education and public sector can make it possible to consider innovative ideas, i.e. what, how, and why certain models will be appreciated by customers according to their changing needs. EdTech therefore has strong potential, followed by financial services and financial technologies (Iivari et al., 2020; Secundo et al., 2021). In the field of technology, cybersecurity can be examined for opportunities for innovation and be a highly profitable business. Covid-19 has accelerated opportunities for digital entrepreneurship from digital diagnosis to virtual care to fitness (Schiavone et al., 2021). Subscription-based online entertainment presents vast opportunities for budding entrepreneurs to harness innovation and creativity to launch new concepts. Covid-19-induced changes in the purchasing and consumption style of individuals have moved organizations to embrace innovative and unique solutions for customer safety as well as their business. Based on these highlights, our study points out “which opportunities are available for digital entrepreneurship and have been advanced during Covid-19, and which related sectors are set to zoom forward with the application of digital technologies and the diffusion of innovations”. 6.2 Implications for practice This study offers implications for digital entrepreneurship not only in emerging areas, but also related areas where DoI helps us to understand how, why, and at what rate new technologies can be adopted and solve a particular issue. Before preparing a business plan in digital space every entrepreneur should (i) evaluate the technological future and associated industry inclination (ii) evaluate the capability of innovation in terms of ease of use and problem solving (iii) asses the required degree of resources to develop and expand the concept in the market and (iv) examine the applicable legal framework and required cybersecurity. Moreover, aspiring and working entrepreneurs in the digital field need to have technical knowledge of the platform/technology they are going to invest in. To set up a digital business, multiple stakeholders need to be integrated (Secundo et al., 2021). For example, in healthcare, digital entrepreneurs have to think about seamless communication and innovative solutions by integrating clinicians, nurses, providers, aggregators, insurance agents and patients. Digital platforms help businesses run smoothly and fulfill the essential needs of customers spending most of their time at home during Covid-19. Technologies ranging from artificial intelligence to big data analytics, cloud computing, augmented reality, educational technology platforms, cybersecurity, and virtual reality present a plethora of opportunities for digital entrepreneurship in a sustainable way and address challenges posed due to Covid-19. Today, with minimal investment like single desktop or server, a start-up can be started in the digital field that requires a sincere effort from an individual or team to grow into a big company in the next few years. Covid-19 resulted in the emergence of digital tools that will shape the world permanently to address the situation and enhance businesses survivability. During Covid-19, governments and supporting agencies had come together in order to empower entrepreneurs to develop innovative solutions to address social problems. Digital entrepreneurs are likely to play a key role in digitizing the economy while focusing on shared value creation. The digital platforms and businesses that people relied on during Covid-19 such as online retailing, digital payment, contactless delivery, and subscriptions to live streaming may henceforth become universal and widen the scope for entrepreneurs further. Entrepreneurs may adopt an integral approach where multiple parties can come together to devise an innovative concept to solve issues. Breakthrough concepts in the digital field can also offer reliable and low-cost solutions to many small and medium enterprises in emerging markets and put them in an advantageous position in the post-Covid era. Entrepreneurs embracing digital technologies led business solutions can contribute to recover the economy faster than other traditional businesses. The key interest of any entrepreneur in developing their digital business today is that one need not to be physically present while doing business. With border closures and frequent lockdowns due to multiple waves, consumer traffic was severely limited in the physical marketplace, forcing organizations to switch to providing services remotely. Despite the global downturn, many entrepreneurs have turned Covid-19-induced challenges into opportunities to create, conceptualize, and launch their ventures. Today, the world exists in an age of dynamism, where constant changes occur and that offers opportunities to entrepreneurs. The transition to digital economy was occurring slowly, and then Covid-19 came along and accelerated the process and offered many opportunities for digital innovation. To start a digital venture, aspiring entrepreneurs have to understand the gap between demand and supply and analyze market patterns and associated opportunities. Before investing in a start-up, one has to evaluate the entrepreneur's own skills in the domain, the scope of scalability, and the knowledge of the legal and institutional environment in that business. The possibility of success in a digital venture is greater when innovation is the critical element and relevant experiences are created for consumers and other stakeholders. In an entrepreneurship journey, no matter how much one prepares, there will always be skills and knowledge that one will gain by doing the business. However, one has to plan a few steps ahead so that the business does not end before it could begin. To avoid early failure of a business, one can create a financial plan to cope up with the flow of funds and be sure of what kind of product or services one wants to offer to which market. 6.3 Limitations and scope for future research Digital technologies have brought a new era to entrepreneurship aspirant as well as professionals. Complexity and concerns in the Covid-19 era have resulted in calls for many start-ups to integrate digitization and technology-related concepts to diffuse innovations and meet the needs of customers, stakeholders and businesses. The emerging propositions demonstrate the impact of Covid-19 and result in novel theorizing in digital technology-led entrepreneurship based on concepts, perspectives, and approaches. The diffusion of innovations and conceptualization of a unique concept are critical for the success of digital entrepreneurship but may not always be true. Some slow adopters can develop the same business or service concept and succeed. Therefore, studies can be conducted to evaluate the success of early adopters and early majority type of ventures. Additionally, social norms and varying acceptance standards in a community may discard the application of DoI. The theory also does not state the degree of rapidity in adopting an innovation among multiple stakeholders. The findings of present study indicate recently-emerging fields for digital entrepreneurship and the application of digital technologies in different ways. In the future, studies can investigate the impact of other grounded theories, such as task-technology fit and the resource-based view in pursuing digital entrepreneurship. In the future, studies can be conducted; those can map the emergence of businesses during Covid-19 and their survival in post Covid-19 era. The propositions can be tested through the lens of institutional theories, where entrepreneurs have different types of pressures to perform and innovate continuously. The present study conducted a semi-structured interview process, whereas in the future, a grounded theory-led exploratory study can be conducted to investigate other emerging areas for digital entrepreneurship. Future research can be conducted in understanding the pre-requisites for diffusion of innovation-bound digital entrepreneurship representing different type of adopters of innovation. 7 Conclusion This study explores innovation-led digital entrepreneurship areas with a high potential to flourish through a semi-structured approach. This study utilizes diffusion of innovation theory view as a basis for Covid-19-induced digital entrepreneurship opportunities. This study adopted a three-layer approach (open, axial, and selective) to perform the thematic analysis. After identifying the themes, a classification and establishment of proposals describing critical areas for digital entrepreneurship was carried out. Furthermore, implications for theory and practice have been drawn on the basis of the developed framework. Hence, this study has attempted to answer the current debate regarding the shift from traditional entrepreneurship to digital entrepreneurship during Covid-19 and the role of DoI in this process. In summary, this study offers meaningful insights for aspiring and existing entrepreneurs in the digital space using technologies innovatively. The application of innovation-oriented solutions is meaningful for digital platforms to stay relevant and meet the needs of their customers and business partners. Acknowledgement: The Open Access funding for this article has been provided by the Qatar National Library. Uncited references Janssen et al., 2018. CRediT authorship contribution statement Sachin Modgil: Conceptualization, Project administration, Writing – original draft. Yogesh K. Dwivedi: Supervision, Resources, Writing – review & editing. Nripendra P. Rana: Data curation, Investigation, Visualization. Shivam Gupta: Methodology, Formal analysis, Writing – original draft. Sachin Kamble: Data curation, Validation. Sachin Modgil is an Assistant Professor at IMI Kolkata in India. He has pursued his PhD from NITIE, Mumbai, India in the domain of technology and operations management. His-areas of interest include big data, artificial intelligence, supply chain management, sustainable operations and production management, lean management, operations and strategy control. He has more than nine years of experience including industry and academia. He has conducted several MDPs and FDPs in operations management domain and has published research papers in leading academic journals. Yogesh K. Dwivedi is a Professor of Digital Marketing and Innovation and Founding Director of the Emerging Markets Research centre (EMaRC) at the School of Management, Swansea University, Wales, UK. In addition, he holds a Distinguished Research Professorship at the Symbiosis Institute of Business Management (SIBM), Pune, India. Professor Dwivedi is also currently leading the International Journal of Information Management as its Editor-in-Chief. His-research interests are at the interface of Information Systems (IS) and Marketing, focusing on issues related to consumer adoption and diffusion of emerging digital innovations, digital government, and digital and social media marketing particularly in the context of emerging markets. Professor Dwivedi has published more than 300 articles in a range of leading academic journals and conferences that are widely cited (more than 30 thousand times as per Google Scholar). He was recently named on the annual Highly Cited Researchers™ 2020 list from Clarivate Analytics. Professor Dwivedi is an Associate Editor of the Journal of Business Research, European Journal of Marketing, Government Information Quarterly and International Journal of Electronic Government Research, and Senior Editor of the Journal of Electronic Commerce Research. More information about Professor Dwivedi can be found at: http://www.swansea.ac.uk/staff/som/academic-staff/y.k.dwivedi/. Nripendra P. Rana is a Professor in Marketing at the College of Business and Economics, Qatar University, Doha, Qatar. Prior to joining Qatar University, Professor Rana held academic positions at some of the leading British universities including his full professorial appointments at the School of Management of University of Bradford and the School of Management at Swansea University. His-current research interests focus primarily on adoption and diffusion of emerging ICTs, digital and social media marketing, and the role of artificial intelligence to understand consumer decision-making and behavior. He has published more than 250 research articles in a range of leading academic journals and conferences. He has co-edited five books on digital and social media marketing, emerging markets and supply and operations management. He has also co-edited special issues, organised tracks, mini-tracks and panels in leading conferences. He is the Chief Editor of International Journal of Electronic Government Research and an Associate Editor of International Journal of Information Management. He has been awarded the Highly Cited Researcher for two consecutive years by Clarivate Analytics in the years 2020 and 2021. Shivam Gupta is a Professor at NEOMA Business School, France with a demonstrated history of working in the higher education industry. Skilled in Statistics, Cloud Computing, Big Data Analytics, Artificial Intelligence and Sustainability. Strong education professional with a Doctor of Philosophy (PhD) focused in Cloud Computing and Operations Management from Indian Institute of Technology (IIT) Kanpur. Followed by PhD, postdoctoral research was pursued at Freie Universität Berlin and SUSTech, China. He has completed HDR from University of Montpellier, France. He has published several research papers in reputed journals and has been the recipient of the International Young Scientist Award by the National Natural Science Foundation of China (NSFC) in 2017 and winner of the 2017 Emerald South Asia LIS award. Sachin Kamble is a Professor of Strategy (Operations and Supply Chain Management) at EDHEC Business School, Lille, France. He holds a Ph.D. in Management, MBA in Operations and a bachelor's degree in Mechanical Engineering. Before joining EDHEC worked with National Institute of Industrial Engineering (NITIE) Mumbai, India. Sachin's teaching and research interests include Operations Management, Supply Chain Management, Big Data Analytics, Industry 4.0 and Digital transformation. He has more than 50 authored/co-authored publications in referred international journals. He has done more than 25 consultancy assignments for leading manufacturing and service organizations representing different sectors. He has also designed and executed various executive development programs for senior level executives in the area of operations and supply chain management. Appendix A: Semi-structured interview schedule Question Criteria considered while designing questions Duration Reference 1. The interaction and dialog is critical aspect of present education system worldwide, be it regular class room or competitive examination preparation or tuition classes. Being country shut, many solutions have been witnessed to keep education imparted. According to you what technology offer towards digital entrepreneurship in a critical sector such as education? 1. Questions are open ended to avoid any acquiescence bias. 2. Questions were designed that allows respondents accepted no matter what the answer is. 3. Question are designed to engage the respondent during the interview schedule. 4. Questions designed carefully to keep general questions first leading to specific ones. 5. Avoided the questions that can prompt to respondents to respond in a particular way or in favor. 30 to 45 min for each interview Leech (2002); McCracken (1988) 2. In a developing country like India, many businesses work primarily in a physical mode and make products available through local “grocery stores”, or “services” through physical visits. In your opinion, how emerging technologies are offering innovative and unique solutions? 3. In pre-Covid world, most of the businesses used to run directly from a set location of a business premises and employees use to operate from their office and factories. Now being most of things from work from home (WFH), what kind of opportunities do you see for digital world? 4. In the Covid-19 era, the most important criteria for consumers when buying a product are its safety, security and sanitization. How technologies in this regard and create opportunities for digital entrepreneurship? 5. Markets being shut for almost two years; many people are missing the touch and feel of buying products such as clothing and other fashion items. Therefore, customer are turning up to E-commerce to fulfill their needs of shopping. How and what kind of technologies offer the opportunities for digital entrepreneurship here? 6. The hospitals and healthcare industry is having high load of service due to its physical mode of operations. There are very few platforms are operating in digitizing the healthcare. What do you feel about expansion of digital healthcare in near future? 7. Many people are health conscious and follow a routine gym exercise through a physical trainer available at facility. Being lockdown and ban on gym opening in most parts of the country, what can be the solutions going forward and sustaining the business? 8. Entertainment is integral part of life and Covid-19 has brought it to standstill with cinema hall closures for a year now. How entertainment industry does is coping up with this situation? ==== Refs References Abubakre, M., Faik, I., & Mkansi, M. (2021). Digital entrepreneurship and indigenous value systems: an Ubuntu perspective. Inf. Syst. J. DOI: 10.1111/isj.12343. Algharabat R. Alalwan A.A. Rana N.P. Dwivedi Y.K. Three dimensional product presentation quality antecedents and their consequences for online retailers: the moderating role of virtual product experience J. Retail. Consum. Serv. 36 2017 203 217 Amin, K.P., Griffiths, M.D., & Dsouza, D.D. (2020). Online gaming during the COVID-19 pandemic in India: strategies for work-life balance. Int. J. Ment. Health Addict. DOI: 10.1007/s11469-020-00358-1. Bacq S. Geoghegan W. Josefy M. Stevenson R. Williams T.A. The COVID-19 Virtual Idea Blitz: marshaling social entrepreneurship to rapidly respond to urgent grand challenges Bus. Horiz. 63 6 2020 705 723 32398883 Bagcchi S. The world’s largest COVID-19 vaccination campaign. The Lancet Infect. Dis. 21 3 2021 322 323 Bai, C., Quayson, M., & Sarkis, J. (2021). COVID-19 pandemic digitization lessons for sustainable development of micro-and small-enterprises. Sustainable Production and Consumption. DOI: 10.1016/j.spc.2021.04.035. Barrett M. Davidson E. Prabhu J. Vargo S.L. Service innovation in the digital age MIS Q. 39 1 2015 135 154 Bentlage E. Ammar A. How D. Ahmed M. Trabelsi K. Chtourou H. Brach M. Practical recommendations for maintaining active lifestyle during the COVID-19 pandemic: a systematic literature review Int. J. Environ. Res. Public Health 17 17 2020 1 22 Brem, A., Viardot, E., & Nylund, P.A. (2021). Implications of the coronavirus (COVID-19) outbreak for innovation: which technologies will improve our lives?Technol. Forecast. Soc. Change. DOI: 10.1016/j.techfore.2020.120451. Cao Z. Shi X. A systematic literature review of entrepreneurial ecosystems in advanced and emerging economies Small Bus. Econ. 57 1 2021 75 110 Carter L. Weerakkody V. Phillips B. Dwivedi Y.K. Citizen adoption of e-government services: exploring citizen perceptions of online services in the United States and United Kingdom Inf. Syst. Manag. 33 2 2016 124 140 Datta, P., & Nwankpa, J.K. (2021). Digital transformation and the COVID-19 crisis continuity planning. J. Inf. Technol. Teach. Cases. DOI: 10.1177/2043886921994821. De, R., Pandey, N., & Pal, A. (2020). Impact of digital surge during Covid-19 pandemic: a viewpoint on research and practice. Int. J. Inf. Manag. DOI: 10.1016/j.ijinfomgt.2020.102171. Dwivedi, Y.K., Ismagilova, E., Hughes, D.L., Carlson, J., Filieri, R., Jacobson, J., & Wang, Y. (2021). Setting the future of digital and social media marketing research: perspectives and research propositions. Int. J. Inf. Manag. DOI: 10.1016/j.ijinfomgt.2020.102168. Dwivedi, Y.K., Hughes, D.L., Coombs, C., Constantiou, I., Duan, Y., Edwards, J.S., & Upadhyay, N. (2020). Impact of COVID-19 pandemic on information management research and practice: transforming education, work and life. Int. J. Inf. Manag. DOI: 10.1016/j.ijinfomgt.2020.102211. Elia, G., Margherita, A., & Passiante, G. (2020). Digital entrepreneurship ecosystem: how digital technologies and collective intelligence are reshaping the entrepreneurial process. Technol. Forecast. Soc. Change. DOI: 10.1016/j.techfore.2019.119791. ET COVID-19 to Accelerate Digital Adoption in India; Jio helped Spur Internet Usage 2020 Morgan Stanley Available at- https://economictimes.indiatimes.com/tech/internet/covid-19-to-accelerate-digital-adoption-in-india-jio-helped-spur-internet-usage-morgan-stanley/articleshow/76031939.cms?utm_source=contentofinterest&utm_medium=text&utm_campaign=cppst Accessed on 27th Feb, 2021 ET (2020b). India Has Over 500 Mn Active Internet users, 14% of 5-11 yrs: IAMAI. Available on- https://economictimes.indiatimes.com/tech/internet/india-has-over-500-mn-active-internet-users-14-of-5-11-yrs-iamai/articleshow/75556305.cms?from=mdr (Accessed on 1st Nov, 2021). Iivari, N., Sharma, S., & Ventä-Olkkonen, L. (2020). Digital transformation of everyday life–How COVID-19 pandemic transformed the basic education of the young generation and why information management research should care?Int. J. Inf. Manag. DOI: 10.1016/j.ijinfomgt.2020.102183. Islam, N., Marinakis, Y., Majadillas, M.A., Fink, M., & Walsh, S.T. (2020). Here there be dragons, a pre-roadmap construct for IoT service infrastructure. Technol. Forecast. Soc. Change. DOI: 10.1016/j.techfore.2017.09.016. Jafari-Sadeghi V. Garcia-Perez A. Candelo E. Couturier J. Exploring the impact of digital transformation on technology entrepreneurship and technological market expansion: the role of technology readiness, exploration and exploitation J. Bus. Res. 124 2021 100 111 Janssen M. Rana N.P. Slade E.L. Dwivedi Y.K. Trustworthiness of digital government services: deriving a comprehensive theory through interpretive structural modelling Public Manag. Rev. 20 5 2018 647 671 Jnr B.A. Use of telemedicine and virtual care for remote treatment in response to COVID-19 pandemic J. Med. Syst. 44 7 2020 1 9 Johnston L.A. World trade, e-commerce, and COVID-19 China Rev. 21 2 2021 65 86 Kamble, S.S., Gunasekaran, A., Kumar, V., Belhadi, A., & Foropon, C. (2021). A machine learning based approach for predicting blockchain adoption in supply Chain. Technol. Forecast. Soc. Change. DOI: 10.1016/j.techfore.2020.120465. Kapoor K. Bigdeli A.Z. Dwivedi Y.K. Schroeder A. Beltagui A. Baines T. A socio-technical view of platform ecosystems: systematic review and research agenda J. Bus. Res. 128 2021 94 108 Kim R.H. Gaukler G.M. Lee C.W. Improving healthcare quality: a technological and managerial innovation perspective Technol. Forecast. Soc. Change 113 2016 373 378 KPMG (2020). The future of higher education in a disruptive world. Available on-https://assets.kpmg/content/dam/kpmg/xx/pdf/2020/10/future-of-higher-education.pdf (Accessed on 14th May 2021). Lallie, H.S., Shepherd, L.A., Nurse, J.R., Erola, A., Epiphaniou, G., Maple, C., & Bellekens, X. (2021). Cyber security in the age of covid-19: a timeline and analysis of cyber-crime and cyber-attacks during the pandemic. Comput. Security. DOI: 10.1016/j.cose.2021.102248. Leech B.L. Asking questions: techniques for semi structured interviews PS: Polit. Sci. Polit. 35 4 2002 665 668 Liu, H., Li, X., & Wang, S. (2021). A bibliometric analysis of 30 years of platform research: developing the research agenda for platforms, the associated technologies and social impacts. Technol. Forecast. Soc. Change. DOI: 10.1016/j.techfore.2021.120827. Madnani D. Fernandes S. Madnani N. Analysing the impact of COVID-19 on over-the-top media platforms in India Int. J. Pervasive Comput. Commun. 16 5 2020 457 475 Marcati A. Guido G. Peluso A.M. The role of SME entrepreneurs’ innovativeness and personality in the adoption of innovations Res. Policy 37 9 2008 1579 1590 McCracken G. The Long Interview (Vol. 13) 1988 Sage McKinsey and Company (2021). Moving digital health forward: lessons on business building. Available on- https://www.mckinsey.com/industries/pharmaceuticals-and-medical-products/our-insights/moving-digital-health-forward-lessons-on-business-building (Accessed on 9th April 2021). Mint (2019). 8 Startups That Are Using Emerging Technologies to Make Life Easier For Others in Real World. Available on - https://www.livemint.com/technology/tech-news/8-startups-that-are-using-emerging-technologies-to-make-life-easier-1562856663586.html (Accessed on 5th Feb, 2021). Muñoz, P., Naudé, W., Williams, N., Williams, T., & Frías, R. (2020). Reorienting entrepreneurial support infrastructure to tackle a social crisis: a rapid response. J. Bus. Ventur. Insights, DOI: 10.1016/j.jbvi.2020.e00181. Nambisan S. Digital entrepreneurship: toward a digital technology perspective of entrepreneurship Entrep. Theory Pract. 41 6 2017 1029 1055 Papadopoulos T. Baltas K.N. Balta M.E. The use of digital technologies by small and medium enterprises during COVID-19: implications for theory and practice Int. J. Inf. Manag. 55 2020 102192 PwC (2020). Pulling the Future forward: The entertainment and Media Industry Reconfigures Amid Recovery. Available on- https://www.pwc.com/gx/en/entertainment-media/outlook-2020/perspectives.pdf (Accessed on 3rd April 2021). Rachul C. Marcon A.R. Collins B. Caulfield T. COVID-19 and ‘immune boosting’on the internet: a content analysis of Google search results BMJ Open 10 10 2020 e040989 Rana N.P. Dwivedi Y.K. Williams M.D. Weerakkody V. Adoption of online public grievance redressal system in India: toward developing a unified view Comput. Human Behav. 59 2016 265 282 Ratten V. Coronavirus (covid-19) and entrepreneurship: changing life and work landscape J. Small Bus. Entrepreneurship 32 5 2020 503 516 Rogers E. Diffusion of Innovations 1962 The Free Press New York, NY Rogers E.M. Singhal A. Quinlan M.M. Diffusion of Innovations 2014 Routledge 432 448 Rogers E.M. Diffusion of Innovations 1995 Free Press New York Schiavone, F., Mancini, D., Leone, D., & Lavorato, D. (2021). Digital business models and ridesharing for value co-creation in healthcare: a multi-stakeholder ecosystem analysis. Technol. Forecast. Soc. Change. DOI: 10.1016/j.techfore.2021.120647. Secundo, G., Gioconda, M.E.L.E., Del Vecchio, P., Gianluca, E.L.I.A., Margherita, A., & Valentina, N.D.O.U. (2021). Threat or opportunity? A case study of digital-enabled redesign of entrepreneurship education in the COVID-19 emergency. Technol. Forecast. Soc. Change. DOI: 10.1016/j.techfore.2020.120565. Shareef M.A. Dwivedi Y.K. Wright A. Kumar V. Sharma S.K. Rana N.P. Lockdown and sustainability: an effective model of information and communication technology Technol. Forecast. Soc. Change 165 2021 120531 Sharma, S., Singh, G., Sharma, R., Jones, P., Kraus, S., & Dwivedi, Y.K. (2020). Digital health innovation: exploring adoption of COVID-19 digital contact tracing apps. IEEE Trans. Eng. Manag., doi: 10.1109/TEM.2020.3019033. Sion G. Smart city big data analytics: urban technological innovations and the cognitive internet of things Geopolit., Hist. Int. Relat. 11 2 2019 69 75 Song A.K. The digital entrepreneurial ecosystem—a critique and reconfiguration Small Bus. Econ. 53 3 2019 569 590 Srinivasan A. Venkatraman N. Entrepreneurship in digital platforms: a network-centric view Strateg. Entrepreneurship J. 12 1 2018 54 71 Subramanian, S.V. (2021). India faces a challenge with its mass vaccination efforts. Lancet Global Health. DOI: 10.1016/S2214-109X(21)00260-6. Sussan F. Acs Z.J. The digital entrepreneurial ecosystem Small Bus. Econ. 49 1 2017 55 73 Szalavetz A. Digital transformation–enabling factory economy actors’ entrepreneurial integration in global value chains? Post-Communist Econ. 32 6 2020 771 792 UNCTAD (2021). How COVID-19 Triggered the Digital and E-Commerce Turning Point. Available on- https://unctad.org/news/how-covid-19-triggered-digital-and-e-commerce-turning-point. Vasenska I. Dimitrov P. Koyundzhiyska-Davidkova B. Krastev V. Durana P. Poulaki I. Financial transactions using FINTECH during the covid-19 crisis in Bulgaria Risks 9 3 2021 1 28 Volberda, H.W., Khanagha, S., Baden-Fuller, C., Mihalache, O.R., & Birkinshaw, J. (2021). Strategizing in a digital world: overcoming cognitive barriers, reconfiguring routines and introducing new organizational forms. Long Range Plann. DOI: 10.1016/j.lrp.2021.102110. Wang, C., Teo, T.S., Dwivedi, Y., & Janssen, M. (2021). Mobile services use and citizen satisfaction in government: integrating social benefits and uses and gratifications theory. Inf. Technol. People. DOI: 10.1108/ITP-02-2020-0097. Zahra, S.A. (2021). International entrepreneurship in the post covid world. J. World Bus. DOI: 10.1016/j.jwb.2020.101143. Zajicek H. Meyers A. Digital health entrepreneurship Digital Health 2018 Springer Cham 271 287

0

PMC9754074

NO-CC CODE

2022-12-16 23:26:26

no

Technol Forecast Soc Change. 2022 Feb 3; 175:121415

utf-8

Technol Forecast Soc Change

10.1016/j.techfore.2021.121415

oa_other

==== Front Surgery Surgery Surgery 0039-6060 1532-7361 Elsevier Inc. S0039-6060(20)30734-0 10.1016/j.surg.2020.10.028 Letter to the Editor Transfer of emergency general surgery patients. Could the role of insurance status be underestimated? Chu Quyen D. MD, MBA ∗ White Robert Keith MD Departments of Surgery at LSU Health Sciences Center-Shreveport, Shreveport, LA Gibbs John F. MD Hackensack Meridian School of Medicine, Hackensack, NJ ∗ Corresponding author. 30 11 2020 5 2021 30 11 2020 169 5 12641264 23 10 2020 © 2020 Elsevier Inc. All rights reserved. 2020 Elsevier Inc. Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. ==== Body pmcTo the Editors: We applaud Bruenderman et al1 for their excellent article. The authors found that the common reasons for emergency general surgery transfers were lack of adequate surgical coverage (20%), surgeon discomfort (24%), or hospital limitations (36%). Surprisingly, more than half of the transfers did not require any urgent surgical intervention and, of those that do, the most common procedures performed were the “bread and butter” (ie, laparoscopic cholecystectomy, small bowel resection, and drainage/debridement of skin and soft tissue infections). Although a small bowel resection may require a higher level of acuity, we wondered why the other 2 operations, which are quite ubiquitous among most general surgeons, would require tertiary care. What were the detailed reasons behind such a transfer, given that 75% of all patients did not require emergent surgical intervention? Could it be that the transfer was done because of some other pernicious factor such as the patient’s insurance status? Although Bruenderman et al1 did not find insurance status to be associated with the likelihood of hospital transfer, this could represent a type II statistical error. In their study, Medicaid represented 21% and self-pay was 9%. In a safety-net institution such as ours, these combined rates can be as high as 60%.2 We have personally found that one of the reasons for the transfer of patients from a private hospital to our hospital was patients’ lack of adequate insurance coverage. Patients were often told that they could either be treated at the current hospital and pay the full bill out of pocket or be transferred to our safety-net hospital where the state will cover their medical bills. The reason for such a transfer was often cited as attributable to “patient preference.” In a landmark study published in 1984, Himmelstein and Woolhandler3 reported that the preponderance of transferred patients were uninsured. More recently, Venkatesh et al4 reported that uninsured patients had a 2.41-fold higher odds of being transferred to another facility than privately insured patients. Bruenderman et al1 speaks to the ongoing perniciousness of legitimate/undisputable versus questionable/disputable transfers. Cherry-picking of the “ideal patients” has been a long-term concern within the medical community. Perhaps, the introduction of risk adjustment indices may allow us to codify these high-need transfer populations. This has the potential to move us in the direction of comparing “apples with apples.”5 In their report, Bruenderman et al1 describe some interesting points surrounding the need for transparency and a system approach. An area that remains unclear involves longer distance transfers, transfers that at times bypass nearby available hospitals. Given the current atmosphere of value-based care delivery and payment models, Bruenderman et al1 underscore the need for us to police ourselves or be subjected to external actors. Again, we commend the authors for adding another dimension to this complex issue and have no criticism of their excellent analysis of a vexing problem, but we would like to bring to focus the possibility of economics playing an insidious role. Funding/Support The authors have no funding sources to report. Conflict of interest/Disclosure None of the authors have financial interests or potential conflicts of interest to disclose. ==== Refs References 1 Bruenderman E.H. Block S.B. Kehdy F.J. An evaluation of emergency general surgery transfers and a call for standardization of practices Surgery 169 2021 567 572 33012562 2 Chu Q.D. Smith M.H. Williams M. Race/ethnicity has no effect on outcome for breast cancer patients treated at an academic center with a public hospital Cancer Epidemiol Biomarkers Prev 18 2009 2157 2161 19622718 3 Himmelstein D.U. Woolhandler S. Harnly M. Patient transfers: medical practice as social triage Am J Public Health 74 1984 494 497 6711726 4 Venkatesh A.K. Chou S.C. Li S.X. Association between insurance status and access to hospital care in emergency department disposition JAMA Intern Med 179 2019 686 693 30933243 5 Kautter J. Pope G.C. Keenan P. Affordable Care Act risk adjustment: overview, context, and challenges Medicare Medicaid Res Rev 2014 10.5600/mmrr.004.03.a02 Accessed October 19, 2020

33272609

PMC9754077

NO-CC CODE

2022-12-16 23:26:26

no

Surgery. 2021 May 30; 169(5):1264

utf-8

Surgery

10.1016/j.surg.2020.10.028

oa_other

==== Front Sci Total Environ Sci Total Environ The Science of the Total Environment 0048-9697 1879-1026 Elsevier B.V. S0048-9697(20)36717-6 10.1016/j.scitotenv.2020.143187 143187 Article Response to ‘Letter to the editor regarding Rodrigues et al. 2020: Is COVID-19 halting wildfires in the Mediterranean? Insights for wildfire science under a pandemic context’ Rodrigues Marcos ⁎ Gelabert Pere J. Ameztegui Aitor Coll Lluís Vega-García Cristina Department of Agricultural and Forest Engineering, University of Lleida, Alcalde Rovira Roure 191, 25198 Lleida, Spain Joint Research Unit CTFC-Agrotecnio, Ctra, Sant Llorenç de Morunys, km 2, 25280 Solsona, Lleida, Spain ⁎ Corresponding author at: Department of Agricultural and Forest Engineering, University of Lleida, Alcalde Rovira Roure 191, 25198 Lleida, Spain. 11 11 2020 20 4 2021 11 11 2020 766 143187143187 © 2020 Elsevier B.V. All rights reserved. 2020 Elsevier B.V. Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. ==== Body pmcIn our work, we presented a preliminary analysis of the potential impacts in wildfires of the societal and public health response to COVID-19 during the lockdown period (Rodrigues et al., 2020). We focused on the winter-spring period, when the lockdown was decreed, and on the Mediterranean region, where human activity is known to be responsible for most of fire activity (Costafreda-Aumedes et al., 2018). Note that at the time we conducted the analysis no information was available about fire activity in summer months so our findings must be properly framed in the March–May period. In that regard, the commentary by Resco de Dios refers to an ‘online first’ version of the manuscript, but in the proof editing process we already suggested a slight modification of the title to help framing our work “Has COVID-19 halted winter-spring wildfires in the Mediterranean? Insights for wildfire science under a pandemic context”. The question addressed in our Short Communication is not whether the year 2020 had higher or lower fire activity. It is to what extent lockdowns and curfews could halt (in the sense of interrupting) wildfire activity when in place, but warning about the forthcoming undesired effects of concatenating successive seasons with reduced fire activity (regardless it comes from lack of ignition sources or unfavorable environmental conditions for fire ignition and spread). That being said, the main points Resco de Dios is raising relate to summer fire activity so we find his answer complementary to our results but cannot see it as a replica, as it does not relate to our analyses or statements. In his commentary, Resco de Dios stated that we “assumed that the main drivers of burned area in the region are either human activity or drought” and that “assuming that drought is the main driver of fire activity is problematic”. We must stress that in our communication we haven't assumed that, in any case. We presumed that winter-spring fires in the Mediterranean region associate to human activity (Costafreda-Aumedes et al., 2018) and we provided a drought-related index (SPEI) to contextualize the yearly comparisons and address the weather circumstances each season (year) underwent. Several works support the use of SPEI as a proxy for thermal and pluviometric anomaly; and Resco de Dios himself has used SPEI for the same purposes we did in the recent work by Nolan et al. (2020), referring to the 2019–2020 extreme fire wave in Australia. Our main assumption is that human activity is the central driver of winter-spring fires (though i.e. winds and plant physiology play an important role as well), thus the cessation in human activity might lead to lower fire incidence. We are fully aware that fire activity is driven by a combination of factors varying spatially and temporally with varying influence. This is specified in our work in different places: “Since wildfires are triggered by the combination of human ignition sources and environmental factors…”; “Even though other factors might be mediating...” (Rodrigues et al., 2020). Likewise, we feel the correlation analysis conducted in Resco de Dios is overly simplistic. In our work, we reported the dispersion value (standard deviation) of the SPEI as an indicator of the spatial variability of SPEI in the region (see Table 1 ). Thus calculating a correlation coefficient or fitting a model from mean SPEI and burned area on a yearly basis to infer anything is misleading. Moreover, the shape of the relationship it is not necessarily linear, as we believe Resco de Dios assumed in his calculations. In any case, if one would proceed and model the relationship between SPEI and burned area with such a limited number of years, it would be more appropriate log transforming (i.e., log-linear model) the dependent variable (BA) so it is closer to a normal distribution and prompts a non-linear relationship. Under such conditions, the log-linear model attains an R2 of 0.22 and, more important, a significant p-value of 0.049. The low R2 has to be expected, since SPEI is not the only driver of wildfires.Table 1 Summary of fire-weather data March–May in the EUMed region (2003−2020). Green shadowing indicates years similar to 2020 in terms of SPEI6 (±0.1 difference). Blue shadowing identifies years similar to 2020 in terms of burned area (most countries showing below average burned area; Z-Score < 0). Red shadowed cells mark the year 2020. Table 1Source: Rodrigues et al. (2020). On a secondary note, one can argue that including 2020 in these models is not entirely appropriate since we hypothesized that it may be an anomalous record, thus potentially biasing regression outputs. If we take away that observation, R2 raises to 0.37 while the p-value is clearly below 0.05 (p = 0.009). Thus, it is clear the relation between SPEI (as an indicator of drought anomaly) and BA (Fig. 1 ).Fig. 1 Summary of log-linear models between SPEI6 and burned area. Points correspond with years. Green points indicate similar years to 2020 in terms of SPEI6 whereas blue points indicate similar years to 2020 in terms of burned area. The orange point marks the position of 2020. The black solid line shows the relationship profile SPEI6-BA including 2020, while the orange line shows the same relationship excluding 2020. Fig. 1 We respectfully disagree from Resco de Dios “that the number of ignitions in different EUMED regions is independent from burned area”, as stated in the commentary note. See for instance the work by Jiménez-Ruano et al. (2019) were an in-depth comparison among fire features and fire weather was conducted exploring the links with cause and season. The association between the number of ignitions and burned area is there. Besides, Rodrigues et al. (2013) found that trends in burned area and in number of fires do not necessarily match each other, likely due to suppression efforts (Curt and Frejaville, 2017). Moreover, some works even suggest a link between human-related fire occurrence and large fires (Costafreda-Aumedes et al., 2015; Nagy et al., 2018). The EFFIS estimations for March–May reported by Resco de Dios are in line with our findings. According to the same EFFIS estimations for Spain -the country for which we reported the strongest decline in fire activity- it can be clearly observed how the profile of both fire ignitions (Fig. 2 ) and burned area (Fig. 3 ) flattens during the lockdown period in March–May 2020 but the 2008–2019 average kept increasing during the same period. Interestingly enough, during the second wave in Spain in the month of September, when measures against COVID-19 were starting again in some regions, the profile has become flat again in 2020 while the former years' trend is still increasing until October.Fig. 2 Weekly distribution of number of fires in 2020 (orange) and average 2008–2019 (blue) in Spain in March–May. Green bars mark the raw difference between 2020 and the reference period. Shadowed background indicate the length and type of COVID-related measures. Data source: EFFIS and Spanish Ministry of Health. Fig. 2 Fig. 3 Cumulative distribution of burned area in 2020 (orange) and average 2008–2019 (blue) in Spain in March–May. Green bars mark the raw difference between 2020 and the reference period. Shadowed background indicate the length and type of COVID-related measures. Data source: EFFIS and Spanish Ministry of Health. Fig. 3 Finally, Resco de Dios states that “the reason underlying the low burned area in 2020 awaits further testing, the mechanism is more likely related to this year's fire weather and fuel availability than to any COVID-19 related impacts on human activity and ignitions”. Again, this statement seems to refer to the entire season 2020 while our work focused on a very specific lockdown period (March–May). In fact, having inspected the weekly profiles (Fig. 2, Fig. 3) we can reassure that the measures took to fight COVID-19 are having some effect again in late summer, where the 2nd wave of COVID-19 infection struck again in Spain, though further testing is required. Nonetheless, we would like to reinforce our agreement in the fact that other factors govern the extent of fires during the main fire season. Our extended findings suggest that fuel-related and weather features mainly control burned area in summer, but outside the summer fire season anthropic factors that limit ignitions greatly influence the burned area output. Nonetheless, the actual effect of COVID-19 on the summer season remains pending of analysis, since curfews were withdrawn in most European countries and analyzing a single season is not sufficient to draw undeniable conclusions. Therefore, we agree further testing is required, using for instance fire statistics documented by each country or examining specific events related to fuel moisture, winds or episodic heat waves. Declaration of competing interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. ==== Refs References Costafreda-Aumedes S. Cardil A. Molina D. Daniel S. Mavsar R. Vega-Garcia C. Analysis of factors influencing deployment of fire suppression resources in Spain using artificial neural networks iForest - Biogeosciences For 008 2015 e1 e8 10.3832/ifor1329-008 Costafreda-Aumedes S. Vega-Garcia C. Comas C. Improving fire season definition by optimized temporal modelling of daily human-caused ignitions J. Environ. Manag. 217 2018 90 99 10.1016/j.jenvman.2018.03.080 Curt T. Frejaville T. Wildfire policy in Mediterranean France: how far is it efficient and sustainable? Risk Anal. 38 2017 472 488 10.1111/risa.12855 28675517 Jiménez-Ruano A. Rodrigues Mimbrero M. Jolly W.M. de la Riva Fernández J. The role of short-term weather conditions in temporal dynamics of fire regime features in mainland Spain J. Environ. Manag. 241 2019 575 586 10.1016/j.jenvman.2018.09.107 Nagy R.C. Fusco E. Bradley B. Abatzoglou J.T. Balch J. Human-related Ignitions Increase the Number of Large Wildfires Across U.S. Ecoregions. Fire 1 2018 10.3390/fire1010004 Nolan R.H. Boer M.M. Collins L. de Dios V. Clarke H. Jenkins M. Kenny B. Bradstock R.A. Causes and consequences of eastern Australia’s 2019–20 season of mega-fires Glob. Chang. Biol. 26 2020 1039 1041 10.1111/gcb.14987 31916352 Rodrigues M. San Miguel J. Oliveira S. Moreira F. Camia A. An insight into spatial-temporal trends of fire ignitions and burned área in the European Mediterranean countries J. Earth Sci. Eng. 3 2013 497 505 Rodrigues M. Gelabert P.J. Ameztegui A. Coll L. Vega-García C. Has COVID-19 halted winter-spring wildfires in the Mediterranean? Insights for wildfire science under a pandemic context Sci. Total Environ. 142793 2020 10.1016/j.scitotenv.2020.142793

33431201

PMC9754078

NO-CC CODE

2022-12-16 23:26:26

no

Sci Total Environ. 2021 Apr 20; 766:143187

utf-8

Sci Total Environ

10.1016/j.scitotenv.2020.143187

oa_other

==== Front Surgery Surgery Surgery 0039-6060 1532-7361 Elsevier Inc. S0039-6060(21)00221-X 10.1016/j.surg.2021.03.029 Plastics/Wound Healing Patient and surgeon experiences with video visits in plastic surgery–toward a data-informed scheduling triage tool Brown-Johnson Cati G. PhD a∗ Spargo Tavish MPAP, PA-C b Kling Samantha M.R. PhD, RDN a Saliba-Gustafsson Erika A. PhD a Lestoquoy Anna Sophia MPH a Garvert Donn W. MS a Vilendrer Stacie MD, MBA, MS a Winget Marcy PhD a Asch Steven M. MD, MPH a Maggio Paul MD c Nazerali Rahim S. MD, MHS, FACS b a Division of Primary Care and Population Health, Stanford University School of Medicine, Stanford, CA b Division of Plastic and Reconstructive Surgery, Stanford Health Care, Stanford, CA c Department of Surgery, Stanford Health Care, Stanford, CA ∗ Reprint requests: Cati G. Brown-Johnson, PhD, Division of Primary Care and Population Health, Stanford University School of Medicine, Medical School Office Building, 1246 Welch Rd., Stanford, CA. 1 5 2021 8 2021 1 5 2021 170 2 587595 21 3 2021 © 2021 Elsevier Inc. All rights reserved. 2021 Elsevier Inc. Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. Background Coronavirus disease 2019 provided the impetus for unprecedented adoption of telemedicine. This study aimed to understand video visit adoption by plastic surgery providers; and patient and surgeon perceptions about its efficacy, value, accessibility, and long-term viability. A secondary aim was to develop the proposed ‘Triage Tool for Video Visits in Plastic Surgery’ to help determine visit video eligibility. Methods This mixed-methods evaluation assessed provider-level scheduling data from the Division of Plastic and Reconstructive Surgery at Stanford Health Care to quantify telemedicine adoption and semi-structured phone interviews with patients (n = 20) and surgeons (n = 10) to explore stakeholder perspectives on video visits. Results During the 13-week period after the local stay-at-home orders due to coronavirus disease 2019, 21.4% of preoperative visits and 45.5% of postoperative visits were performed via video. Video visits were considered acceptable by patients and surgeons in plastic surgery in terms of quality of care but were limited by the inability to perform a physical examination. Interviewed clinicians reported that long-term viability needs to be centered around technology (eg, connection, video quality, etc) and physical examinations. Our findings informed a proposed triage tool to determine the appropriateness of video visits for individual patients that incorporates visit type, anesthesia, case, surgeon’s role, and patient characteristics. Conclusion Video technology has the potential to facilitate and improve preoperative and postoperative patient care in plastic surgery but the following components are needed: patient education on taking high-quality photos; standardized clinical guidelines for conducting video visits; and an algorithm-assisted triage tool to support scheduling. ==== Body pmcIntroduction The coronavirus disease 2019 (COVID-19) pandemic forced healthcare systems to implement telemedicine as standard practice across service lines, including surgical subspecialties. Telemedicine in surgery has been shown to improve access, reduce healthcare costs, save time, decrease travel burden, and improve quality of care.1, 2, 3, 4, 5, 6, 7 Most studies have focused on the postoperative period and indicate that video visits can be delivered safely, with similar patient outcomes and similar or superior patient and surgeon satisfaction to office visits.2 , 3 , 5 , 8 , 9 The limited studies that have evaluated telemedicine in the preoperative surgical context indicate virtual care can be effective for screening, discussing surgical options, and diagnosing patients.10, 11, 12 A systematic review of telemedicine in plastic surgery found benefits related to postoperative monitoring, access to high-level care in rural communities, and cost.13 Early evidence indicates that applications of telemedicine for triage decisions and postoperative monitoring have patient outcomes comparable to in-patient visits.14, 15, 16, 17 Video visits can also facilitate effective and timely detection and management of postoperative complications.15 , 18 However, concerns remain regarding: surgeons’ ability to accurately assess wounds, such as burn depth19; effective communication when video quality and connectivity are subpar20, 21, 22; and patient security and privacy.18 Furthermore, outside of patient education videos, very little is known about video visits in preoperative plastic surgery settings. The lack of established best practices in plastic surgery telemedicine is another limitation.13 , 23 We report patient and surgeon experiences with video visits during the California COVID-19 stay-at-home order which prompted both cancellation of non-essential surgeries and rapid, full-scale implementation of telemedicine for patient visits. Using a mixed-methods approach, surgeon and patient perceptions were analyzed along with scheduling data to understand the potential and limitations of preoperative and postoperative video visits in plastic surgery. Results were used to inform the development of a scheduling triage tool for appropriate use of video visits in plastic surgery. This quality improvement project was given a non-research determination by the Stanford University Institutional Review Board (#56131) Methods Setting The study was conducted at Stanford Health Care (SHC), a quaternary academic medical center, and ValleyCare, an affiliate clinic within the Division of Plastic Surgery. Video visits were rolled-out system-wide in response to the San Francisco Bay Area and California COVID-19 stay-at-home order in March 2020 and continued through the spring and summer of 2020. This quality improvement project was given a non-research determination by the Stanford University Institutional Review Board (#56131). The vast majority of patients are seen for a consultation visit, and a preoperative visit before surgery (2 presurgery visits). The initial evaluation visit includes discussion of possible surgery, and the preoperative visit is focused on preoperative evaluation. Primary care providers are not routinely involved, unless the patient has a complicated medical history that requires optimization through the perioperative period. Intervention Video visits were rolled-out system-wide in response to the San Francisco Bay Area and California COVID-19 stay-at-home order in March 2020 and continued through the spring and summer of 2020. Video visits were offered through Vidyo (Hackensack, NJ), a video conference software integrated into Epic (Verona, WI), the electronic health record system used at SHC. At the start of the pandemic, all providers were given video-visit-enabled HP Elitebooks (HP Inc. Palo Alto, CA) to enable them to complete video visits from home. Clinic computers were used to complete video visits from the office. In addition, surgeons and other clinic staff were offered a 30-minute training module through HealthStream (Nashville, TN). Patients eligible for video visit consultations were sent prompts via Stanford’s MyHealth app, with instructions on how to sign in for the video visit. The MyHealth help (phone) line was also available if they encountered any technical issues. Qualitative interviews Four qualitatively trained interviewers (E.A.S.G., A.S.L., C.G.B.J., and M. Verano) conducted semi-structured phone interviews scheduled for 30 minutes between April and May 2020 with a convenience sample of 10 plastic surgeons conducting video visits and a stratified random sample of 11 preoperative and 9 postoperative patients. Patients who completed a video visit with a plastic surgeon between April 13, 2020, and April 24, 2020, were eligible. A total of 101 patients met this criterion as identified through managerial reporting tools. Patients were stratified by preoperative versus postoperative visit and randomized by simple random selection without replacement. Patients had participated in either one or the other (preoperative or postoperative video visits) but not both at the time this study was conducted. Interviewed providers represented a wide variety of sub-specialties, including breast reconstruction, hand surgery, oromaxillofacial surgery, and complicated and uncomplicated wounds and burns. Patient surgery types included breast reconstruction, hand and foot surgery, liposuction, lymphedema, oromaxillofacial, and wound/spine (see Table I ).Table I Characteristics of interviewed patients (n = 20) n Age  20–30 3  31–40 5  41–50 2  51–60 7  61–70 2  71–80 0  81–90 0  >90 1 Sex  Male 9  Female 11 Race/ethnicity  White 11  Hispanic 3  Asian 2  Other/unknown 4 Type of surgery consulted for  Breast reconstruction 5  Foot surgery 1  Hand surgery 7  Liposuction 1  Lymphedema 1  Oral and maxillofacial surgery 4  Wound/spine 1 Pre-/postoperative consultation  Preoperative 11  Postoperative 9 New versus follow-up patient visit  New patient visit 4  Follow-up patient visit 16 The number of successful video visits completed on the Epic platform for each interviewed SHC surgeon before their semi-structured interview was determined, and mean and standard deviation are reported. Data were not available for the ValleyCare surgeons. At the time of the interviews, SHC surgeons had successfully completed 16.1 ± 8.9 (range: 5–29) video visits on the Epic platform. Interviews with providers included: questions about satisfaction, acceptability, and comfort with video visits; opportunities and limitations of a video physical exam; potential sustainability/future of video visits in plastic surgery; best practices for video visits; perceptions of video impact on patient-provider connection; and cost and reimbursement for video visits. Patient interviews similarly focused on: patient experience and satisfaction with video visits; impacts of video on patient-provider relationships, trust, patient health, and wellbeing; and the future potential for video visits. Interview guides are available upon request. A multi-phase analysis approach leveraged rapid analytic24 and constant comparison25 approaches to: (1) extract early themes using templated research notes (rapid); (2) compare findings across participants using a matrix analysis (rapid); and (3) determine consensus synthesis of findings with frequent qualitative group and larger team discussions (rapid and constant comparison). C.G.B.J., E.A.S.G., and T.S. condensed themes during weekly meetings. Finally, results were presented to the full team and refined iteratively with additional data examination for final consensus. The main themes that related to barriers and facilitators were further discussed with clinical leaders to develop the proposed Triage Tool for Video Visits in Plastic Surgery. Scheduling data Scheduling data were used to assess video visit adoption. Data were extracted from January 5, 2020, through June 20, 2020, to capture volumes before and after the stay-at-home order (initiated week of March 15, 2020). The first week of the stay-at-home order was considered a transition week and excluded from analysis. The 13-week implementation period was thus March 22, 2020, through June 20, 2020, during which 22 providers (14 plastic surgeons, 5 nurse practitioners, 3 physician assistants) conducted at least 1 video visit and were thus included. Post-implementation scheduling data for all providers were compared to the pre-implementation period. The transition to video visits and adoption was described with 3 outcomes: (1) the number of in-person and video visits conducted per week and per provider from January 5, 2020, through June 20, 2020; (2) the proportion of visits completed via video during the 13-week period for all visit types; and (3) the proportion of visits completed via video during the 13-week period by provider. The third measure is presented as the number of providers conducting 0 to <10%, 10 to <20%, 20 to <30%, 30 to <40%, 40 to <50%, and 50 to <60% of their visits via video. The number of successful video videos completed on the Epic platform for each interviewed SHC surgeon before their semi-structured interview was determined, and mean and standard deviation are reported. Data were not available for the ValleyCare surgeons. All quantitative data were processed and analyzed using SAS (version 9.4; SAS Institute, Inc., Cary, NC) and R (version 4.0.2; RStudio, Boston, MA). Results Video visit utilization Before the stay-at-home orders, video visits were not utilized in plastic surgery. Upon implementation of video visits, however, surgery providers adopted video visits rapidly, although video visit utilization was variable among providers and visit types. Of the 22 providers who conducted video visits, 14 (61%) integrated them into their practice within the first 2 weeks of the stay-at-home order, which increased to 21 (91%) by the seventh implementation week. Video visits were implemented rapidly, with 30–118 video visits/wk (Fig 1 ), and an average of three video visits/wk/provider. The proportion of visits completed via video varied among providers, however; 11 (50%) completed 20%–30% of visits via video whilst 6 (27%) seldom used video (0%–10%) (Fig 2 ).Fig 1 Video and in-person visits completed by plastic surgeons, physician assistants, and nurse practitioners before and after (n = 22) the COVID-19 stay-at-home orders (initiated week of Mar 15, 2020). N.B. The first week of the stay-at-home order was considered a transition week and excluded from analysis. The 13-week implementation period was thus March 22, 2020, through June 20, 2020. COVID-19, coronavirus disease 2019. Fig 2 Proportion (%) of completed visits conducted via video by plastic surgeons, physician assistants, and nurse practitioners during the 13-week implementation period (Mar 22, 2020, through Jun 20, 2020). During the implementation period, 24% of all patient visits were completed via video (Fig 3 A), however, video visit utilization varied by visit type. Video visits were adopted at similar rates for new and return patient visits (25% and 24% of completed visits, respectively) as shown in Figures 3B and 3C. Video visits were utilized less for preoperative (21%; Figure 3, D) than postoperative visits (46%; Figure 3, E).Fig 3 Proportion (%) of visits completed via video and in-person during the 13-week implementation period (Mar 22, 2020, through Jun 20, 2020) for: (A) all patient visits, (B) new patient visits, (C) return patient visits, (D) preoperative patient visits, and (E) postoperative patient visits. Qualitative data: patient and surgeon interviews Patient and surgeon interview themes centered on satisfaction, the virtual physical examination, and patient-surgeon relationship (see Table II for exemplar patient and surgeon quotes).Table II Exemplar quotes from patient and surgeon interviews organized by theme: satisfaction, the virtual physical examination, and the patient-surgeon relationship Themes Exemplar patient quotes Exemplar surgeon quotes Satisfaction Travel “[The post-op video visit] worked for me… To drive that far for…a ten-minute appointment is kind of ridiculous when we could do it through video conferencing.” (Pt4, 36-year-old female patient) “…it's really good to meet patients from far away. That travel is a large burden for them, to assess whether or not they're potential candidates for surgery …… they're not just disappointed that you told them there's nothing you can do, they're disappointed that they're going to take a 12-hour trip for you to say in 10 minutes, "There's nothing I can do for you."” (MD9, plastic surgeon) Access - scheduling “Now it's like, ‘Oh you want to video visit again? Let me know, I can do it tomorrow, next week, whatever…’” (MD1, plastic surgeon) “…once the pandemic is over, I still think that the video visits is going to be my first go to mechanism. … It's faster, it's efficient, we use less resources than clinic……” (MD1, plastic surgeon) The virtual physical examination Patient-assisted virtual physical examination “Maybe, some sort of tutorial for the patient. Yeah or some kind of an app that does that [coaches patients on how to conduct a self-exam].” (MD6, plastic surgeon) “I can tell them to press on their stomach, but are they going to feel a colon tumor? Are they going to know what they're feeling? It's okay if it's symptomatic and it's sore, inflamed or infected. Yeah, they'll notice that. But not painless masses.” (MD5, plastic surgeon) Visualization “…he couldn't view my wounds up close, he had to go by a photo. So, the detail of the video conference wasn't as good as it would be in person.” (Pt1, 52-year-old male patient) “…patients really struggle to give you perspective on the body parts that you're trying to look at during the physical exam...even though they have a good quality video, I'm trying to look at their hand and they're showing me their sofa across the room.” (MD9, plastic surgeon) Tactile exam “Because of the intricate and also intimate nature of it (breast reconstruction) ...… I want to be absolutely sure that every decision that I make is a very informed one, and without understanding my vascularity, my skin integrity, there's just no possible way to determine that over video. A doctor has to be able to assess that and evaluate that firsthand.” (Pt6, 53-year-old female patient) “Plastic surgery is very much a kind of touch-and-feel type of surgery where you need to be able to understand tissue laxity, understand tissue quality, the mobility of the tissues...you can't do that over video.” (MD8, plastic surgeon) The patient-surgeon relationship Connection “I would obviously probably go with someone who I could just meet face-to-face, have more time, see their body language, and have a better understanding of what type of surgeon they are, what type of person they are...you really don't get that from a video exchange.” (Pt1, 52-year old male patient) “A lot of the issues that we're dealing with in reconstructive surgery, it takes a lot of time to counsel patients. I think that there is probably a lack of that personal connection with the patient when you're on the video visit, and so I think there are some challenges there.” (MD10, plastic surgeon) “He explained everything in detail just the same as if I were there.” (Pt 8, 39-year old female patient) Trust “…you just have a more personable experience. ...when I tell him something I don't really know if he gets it. I can't see the expression on his face, I can't get a read on what he's thinking ... if it was in person, I could see the expression on his face. I could gain more feedback depending on what that expression is...it's a whole physical thing that you'll never be able to reproduce over the phone.” (Pt11, 58-year-old male patient) Patient and surgeon semi-structured interview guides assessed participant experience of preoperative and postoperative virtual care, perceived limitations of care received via video, and comfort level with surgeries preceded via telemedicine exclusively. Interviews were either (1) audio recorded following verbal consent, or (2) when audio recording was not preferred, interviewers took detailed notes of the conversation for subsequent analysis. Authors E.A.S.G., A.S.L., and M. Verano condensed interview notes into structured research notes based on a template developed with qualitative expert C.G. Brown-Johnson. To enable cross-participant and cross-role (patient, provider) comparison, concepts of interest from the templated research notes were transferred into an analytic matrix where rows represented participants, and columns represented themes. Three themes were identified a priori: (1) satisfaction with video visits; (2) the virtual physical examination; and (3) the patient-provider relationship. For the matrix analysis, excerpts from transcripts were reviewed by theme to refine findings and allow for comparison across all interviewed participants. Satisfaction Overall, patients were satisfied with video visits with their surgeon, however surgeons’ perceptions were mixed. Patients’ and surgeons’ satisfaction with video visits were related to reduced patient travel and increased access due to ease of scheduling and timing. Barriers to satisfaction revolved around technological inefficiencies, negative impacts of time pressure, and patient distractions. Reduced patient travel and time were clear benefits of video visits reported by patients and surgeons. Many patients suggested video visits should be standard practice given that travel is not required for video. Some patients also noted not needing to take time off work as a benefit of video visits. Video visits were also easier and more convenient to schedule than in-person visits. Patients appreciated not having to wait in the clinic and were more tolerant of surgeons running late since they were home. In the unusual case that a surgeon was running late, patients suggested that notification of delays could further improve their experience. Surgeons positively reported that video visits were much quicker without the need for rooming. Surgeons also reported that increased access and fewer communication barriers ultimately improved continuity-of-care. Barriers to satisfaction included technical delays and directing physical examinations over video, which led to longer consultation times. Notably, 1 surgeon felt less equipped to manage the anxiety that accompanies cancer patients due to time constraints in a video setting, “… [it] take[s] a lot of time to counsel patients” (MD10, plastic surgeon). A minority of surgeons also found that distractions from pets and children during video visits sometimes hindered information exchange. The virtual physical examination Several patients and surgeons agreed that the in-person physical examination could not easily be replicated via video and that conducting an examination over video was challenging. On the positive side, many patients believed that the quality of care and medical advice they received were similar to in-person visits despite the limitations of a virtual format. Some patients who reported not needing an extensive physical examination were even able to successfully navigate a virtual physical examination that was conducted by the patient as directed by the surgeon (ie, a “Patient-Assisted Virtual Physical Exam”26). Several patients who underwent a patient-assisted virtual physical examination were comfortable with their experience, even when the examination included private areas of their body. Surgeons, however, were less comfortable with patient self-examinations. One surgeon expressed that it would be awkward to perform sensitive examinations in front of the camera. Directing patients to perform physical examinations, particularly those that entailed performing maneuvers, were challenging and some surgeons felt that they could not trust patients to successfully identify complicating issues, such as tumors. Not surprisingly, patients and surgeons alike expressed that the main limiting factor for quality of care was related to the physical examination. Virtual physical examinations were limited in 3 aspects: (1) visual observation, (2) tactile examination, and (3) comfort with following the surgeons’ directions to examine themselves and communicate those findings back to the surgeon. Patients and surgeons agreed that video quality, camera angle, perspective, and the device used (eg, phone versus tablet versus computer) affected the ability to successfully visualize a surgical site or body part during the physical examination. Poor video quality rendered inadequate visualization of surgical areas, limiting surgeons’ ability to acquire the necessary information through observation. Even when video quality was high, scars were “almost impossible” to see over video, in some situations due to patients’ lack of expertise with framing and focusing the video on body parts of surgical interest. Surgeons suggested that patients receive more education on the technology to prepare them for video visits including ways to visualize certain body parts/surgical sites over video, and tips on video stabilization and correct lighting. Surgeons preferred that patients take the video call on a tablet or computer to ensure camera stability. By coaching patients to use better lightning or optimize angles, sometimes surgeons were able to better visualize the area, although this was still challenging. Surgeons also mentioned that standardized photos were required in many instances for documentation and insurance purposes; therefore, it was not ideal that patients take their own photos due to lighting issues. One surgeon suggested patient education including an outline to help position the body part/surgical site when taking a photo outside of the clinical setting. The surgeon’s inability to gather tactile information (eg, assess skin quality, wounds, and lesions, and feel for masses) and view possible donor tissue sites—healthy areas of the body that might serve as donor tissue to be transplanted to target areas—were considered major limitations of video visits by both patients and surgeons. In addition, the inability to carry out a heart and lung examination and take specific patient measurements were also considered limiting factors that could create problems during surgery. Patients with a recent physical examination, testing, and/or imaging before the video visit, however, made preoperative video visits more acceptable to some surgeons. Surgeons noted that hand surgeries were particularly challenging cases, where determining range of motion and assessing sensation and point tenderness over video was very difficult. Visualization problems were most challenging for patients with hand injuries and those who did not have nearby family/caregivers to assist with the video visit. The patient–surgeon relationship Video visits impacted the patient-surgeon relationship in 2 categories: (1) connection and (2) trust. Patients reported establishing trusting relationships with consulting surgeons over video. In contrast, some surgeons noted having a harder time developing a personal connection with patients over video. Most patients reported that their interaction was the same as an in-person visit since surgeons took their time providing information and responding to their queries. Patients rarely noted that communication and establishing a relationship with their surgeon was harder over video; however, when it was mentioned, this was cited as a reason to prefer an in-person visit. Few surgeons discussed the patient-surgeon relationship, but those who did reported it challenging to develop a personal connection with patients over video: “…you do lose some of the nuances and subtleties of in-person interaction” (MD1, plastic surgeon). Patients and surgeons exhibited divided and varying levels of comfort undergoing surgery without first having an in-person consultation. Where patients expressed comfort with a direct video-to-surgery trajectory, trust in the surgeon was paramount. Although patients had split views on whether they would allow a surgeon to perform surgery on them without an in-person consultation, patients uniformly trusted their surgeons, partly owing to the institution’s good reputation. Similarly, surgeons’ views on performing surgery without first seeing the patient in person were divided. The limiting factor was the inability to perform a complete preoperative physical examination oneself and to determine whether a patient was a good candidate for surgery. One surgeon reflected, “I would never do an operation based on a video call” (MD8, plastic surgeon). Some surgeons, however, identified specific situations where they would proceed with an operation after a preoperative video visit without an in-person visit. For instance, referral of a patient by another surgeon would increase confidence in decision-making. Additionally, surgeons felt more comfortable not having an in-person preoperative visit if they had already operated on the patient (eg, for second-stage surgery) or had an established relationship. Surgeons comfortable with the preoperative virtual visits expressed that, before operating, they confirm that the patient is comfortable with the process and assure them that they would meet face-to-face on the day of surgery. The proposed triage tool for video visits in plastic surgery Patient and surgeon perceptions of future use and acceptability of video visits varied based on the goal of the visit and formed the initial basis for the proposed Triage Tool for Video Visits in Plastic Surgery (Fig 4 ). In the preoperative setting, interviewed patients and surgeons agreed that informational visits were ideal for video. Surgeons also saw use for telemedicine for initial screening of patients to potentially rule out surgery, but video visits were not considered suitable to determine what type of procedure was needed. Other preoperative scenarios preferred for video visits were non-emergency informational visits, patients traveling long distances, standard/simple or co-surgeon cases, and/or lower-risk patients.Fig 4 Data-informed tool to assist triage for video visits in plastic surgery. N.B. Informed by semi-structured interviews with video visit patients (n = 20) and surgeons (n = 10) who conducted 21.4% of preoperative and 45.5% of postoperative visits via video during local COVID-19 stay-at-home orders; ∗If access is otherwise impeded; †Outside of system/ known providers; ‡Smoking, smokeless tobacco and vaping included. COVID-19, coronavirus disease 2019. Postoperative video visits were acceptable as a future state for most surgeons and generally acceptable to patients although not preferred. Patients expressed that postoperative, long-term, and follow-up care beyond the acute recovery stage were the ideal for video visits, but immediate postoperative appointments by video were less acceptable. Surgeons agreed that uncomplicated postoperative visits and long-term follow-up appointments were appropriate future uses for video visits, unless procedures like suture or drain removal are necessary (Fig 4). Patients indicated that video visits were not acceptable after an acute injury or complication, especially if a physical examination or additional testing was required. On the other hand, check-ups with patients in skilled nursing facilities, where transport could be expensive and challenging, would be ideal in the future for video format. Finally, surgeons expressed a need for clearer guidance on how to prioritize video visit utilization in their clinic to support their work and ensure that patient expectations and needs are met. Discussion This mixed-methods study is one of the first to examine plastic surgery patients’ and surgeons’ attitudes toward video visits and document adoption across plastic surgery providers and visit types during local COVID-19 stay-at-home orders. Video visits were rapidly adopted by providers, which allowed patients to receive the necessary care while minimizing the risk for in-person pathogen exposure. It is telling that despite universal utilization in this extreme situation, providers reported different levels of comfort with the incorporation of telemedicine visits into their practices, and different providers reported planning to incorporate telemedicine differently into their ongoing practice. Future plans ranged from never utilizing telemedicine going forward to planning to use telemedicine preferentially unless contraindicated. Understanding the patient and surgeon experience and satisfaction with video visits expands existing literature and helps identify specific needs of plastic surgery video visits in preoperative and postoperative settings. Both patients and surgeons reported positive experiences but also limitations, including the virtual physical examination. Video visits were found to be satisfactory and acceptable for lower-risk situations, as has been documented in previous research.27 Indeed, surgeons and patients agreed that the best future use of video visits would be in lower-risk informational preoperative visits and long-term postoperative visits. Most appropriate for video telemedicine would be postoperative visits where the patients do not feel they are experiencing any problem, and where nothing specific like stitch removal, drain removal, or tissue expansion is needed. Conversely, in a typical practice model with 2 preoperative visits, most providers concur that conducting both by video would not be comfortable except when the patient is already a patient of the provider or when the patient is referred by a very trusted colleague. The primary reported driver for patient acceptance was trust in their surgeon; patients relied on their surgeon to determine whether a video visit would be appropriate for them and to provide expert care via video. Patient trust in surgeons as a driver for satisfaction has been well-documented.28, 29, 30, 31 Patients value their surgeon’s interpersonal skills as they develop trust, which was also observed in this study where trust was recognized as an important aspect of virtual care delivery. However, trust (or lack of trust) may be less binary and more nuanced; for example, a patient that trusts a surgeon due to the reputation of the institution could have that trust evaporate in the face of complications, especially if personal relationships had not been adequately established. The virtual physical examination continued to be a challenge, as is the case in other specialties.32 However, our data identify opportunities to address limitations of video quality and the physical exam. In the evaluation clinic, patient education instructions were created to instruct patients on how to take high-quality photos. For example, patients having breast reduction surgery received a set of instructions regarding simple lighting and positioning to help direct taking high-quality photos of their breasts that could be uploaded to the electronic medical record. Locally successful, photos resulting from these directions were submitted to insurance companies and approved for their respective surgeries. Beyond the logistics of photo documentation, 3 previously identified pillars of positive patient expectation-setting include patient education, patient acceptance, and patient-provider trust.33 A multi-dimensional approach addressing each of these pillars may be key to long-term video visit acceptability. Our evaluation suggests that video visits are accepted by patients and do not harm trust; efforts to support patient satisfaction may be enhanced with additional patient education. From an operational perspective, our findings suggest a rationale for using video visits primarily for postoperative care; once a surgeon has established a trusting relationship during an in-person consultation, a postoperative video visit leverages that trust. However, our study’s implementation rates suggest that surgery providers vary widely in their readiness to integrate video visits. Individual surgeons may vary in their surgical approach.34 As specialist practices, surgery clinics often have detailed documentation of individual order-of-operation preferences per surgeon. In the study clinic, after this evaluation, these preferences were expanded to include detailed documentation of surgeon preferences for video/in-person visits, based on ICD-10 diagnosis codes. We anticipate that this information will help scheduling and clinic flow and may be highly relevant in the current context of an academic institution—a setting that may be less prescriptive than managed-care organizations. Surgeon preferences can also be mapped to the guidelines outlined in the scheduling triage tool for identifying video-appropriate patient candidates, according to visit type, case, surgeon’s role, and patient characteristics (Fig 4). Anecdotally, clinic schedulers struggled to identify appropriate candidates for telemedicine without the scheduling triage tool; the factors in the triage tool could be used to build decision tools, algorithms, or flags to assist schedulers in identifying patients appropriate for telemedicine. The factors in the triage tool can also serve as the foundation for development and standardization of an algorithm to ultimately support the development of automated procedures to support telemedicine/non-telemedicine triage. Indeed, algorithm systems and Information Technology approaches to scheduling have been successfully piloted to schedule surgeries in the operating room.35 Strengths and limitations The COVID-19 stay-at-home order and its impact on care delivery provided a unique opportunity to rapidly evaluate the widespread use of telemedicine in plastic surgery but may limit the applicability of findings in a post–stay-at-home period. Our study included patients and providers from only 2 institutions; patient and surgeon perspective generalizability may be limited. While physician and patient perspectives on the impact of video visits on quality are important to understand, this study did not assess the impact of video visits on processes or outcomes of plastic surgery encounters. While encounter data with physician assistants and nurse practitioners were included in our quantitative analysis, resource constraints limited our ability to capture interviews from this population, thereby limiting our understanding of how this integral role shapes video visit dynamics. Finally, other long-term impacts of video visits, such as increased utilization of imaging, laboratory studies, and in-person visits, were not evaluated. In conclusion, video visits were viewed as acceptable by patients and surgeons in plastic surgery in terms of quality of care but were limited by the inability to perform a physical examination. Pre–video-visit patient education may be a potential solution to partially overcome this limitation and improve the virtual physical examination. Additionally, our results informed development of a triage tool to determine the appropriateness of video visits for individual patients. Before widespread dissemination or permanent adoption, however, further study is needed to understand the appropriate use of video visits in the post-pandemic period as well as their impact on patient outcomes and healthcare utilization. Funding/Support This quality improvement project evaluation was funded by Stanford Health Care ICDP and Stanford Division of Primary Care and Population Health. Conflict of interest statement Authors have no conflicts of interest. Acknowledgments We would like to thank our participants, and also Mae Richelle Verano for her contribution to data collection. Cati G. Brown-Johnson and Tavish Spargo are equal contributing first authors. ==== Refs References 1 Asiri A. AlBishi S. AlMadani W. ElMetwally A. Househ M. The use of telemedicine in surgical care: a systematic review Acta Inform Med 26 2018 201 206 30515013 2 Soegaard Ballester J.M. Scott M.F. Owei L. Neylan C. Hanson C.W. Morris J.B. Patient preference for time-saving telehealth postoperative visits after routine surgery in an urban setting Surgery 163 2018 672 679 29398042 3 Gunter R.L. Chouinard S. Fernandes-Taylor S. Wiseman J.T. Clarkson S. Bennett K. Current use of telemedicine for post-discharge surgical care: a systematic review J Am Coll Surg 222 2016 915 927 27016900 4 Tel A. Bortuzzo F. Pascolo P. Costa F. Sembronio S. Bresadola V. Maxillofacial surgery 5.0: a new paradigm in telemedicine for distance surgery, remote assistance and webinar Minerva Stomatol 69 2020 191 202 32203647 5 Hwa K. Wren S.M. Telehealth follow-up in lieu of postoperative clinic visit for ambulatory surgery: results of a pilot program JAMA Surg 148 2013 823 827 23842982 6 Gardiner S. Hartzell T.L. Telemedicine and plastic surgery: a review of its applications, limitations and legal pitfalls J Plast Reconstr Aesthet Surg 65 2012 e47 e53 22178033 7 Urquhart A.C. Antoniotti N.M. Berg R.L. Telemedicine--an efficient and cost-effective approach in parathyroid surgery Laryngoscope 121 2011 1422 1425 21647908 8 Cremades M. Ferret G. Parés D. Navinés J. Espin F. Pardo F. Telemedicine to follow patients in a general surgery department. A randomized controlled trial Am J Surg 219 2020 882 887 32252983 9 Yoder L.H. McFall D.C. Cancio L.C. Use of the videophone to collect quality of life data from burn patients Int J Burns Trauma 2 2012 135 144 23272295 10 Wood E.W. Strauss R.A. Janus C. Carrico C.K. The use of telemedicine in oral and maxillofacial surgery J Oral Maxillofac Surg 74 2016 719 728 26707428 11 Sudan R. Salter M. Lynch T. Jacobs D.O. Bariatric surgery using a network and teleconferencing to serve remote patients in the Veterans Administration Health Care System: feasibility and results Am J Surg 202 2011 71 76 21601825 12 Lee S. Broderick T.J. Haynes J. Bagwell C. Doarn C.R. Merrell R.C. The role of low-bandwidth telemedicine in surgical prescreening J Pediatr Surg 38 2003 1281 1283 14523806 13 Vyas K.S. Hambrick H.R. Shakir A. Morrison S.D. Tran D.C. Pearson K. A systematic review of the use of telemedicine in plastic and reconstructive surgery and dermatology Ann Plast Surg 78 2017 736 768 28328635 14 Kiranantawat K. Sitpahul N. Taeprasartsit P. Constantinides J. Kruavit A. Srimuninnimit V. The first Smartphone application for microsurgery monitoring: SilpaRamanitor Plast Reconstr Surg 134 2014 130 139 25028822 15 Hee Hwang J. Mun G.-H. An evolution of communication in postoperative free flap monitoring: using a smartphone and mobile messenger application Plast Reconstr Surg 130 2012 125 129 22743879 16 Engel H. Huang J.J. Tsao C.K. Lin C.-Y. Chou P.-Y. Brey E.M. Remote real-time monitoring of free flaps via smartphone photography and 3G wireless Internet: a prospective study evidencing diagnostic accuracy Microsurgery 31 2011 589 595 22072583 17 Trovato M.J. Scholer A.J. Vallejo E. Buncke G.M. Granick M.S. eConsultation in plastic and reconstructive surgery Eplasty 11 2011 e48 22140594 18 Pozza E.D. D’Souza G.F. DeLeonibus A. Fabiani B. Gharb B.B. Zins J.E. Patient satisfaction with an early smartphone-based cosmetic surgery postoperative follow-up Aesthet Surg J 38 2017 101 109 29117293 19 Hop M.J. Moues C.M. Bogomolova K. Nieuwenhuis M.K. Oen I.M.M.H. Middelkoop E. Photographic assessment of burn size and depth: reliability and validity J Wound Care 23 2014 144 152 24633060 20 Syed-Abdul S. Scholl J. Chen C.C. Santos M.D.P.S. Jian W.-S. Liou D.-M. Telemedicine utilization to support the management of the burns treatment involving patient pathways in both developed and developing countries: a case study J Burn Care Res 33 2012 e207 e212 22249104 21 Whitehead E. Dorfman V. Tremper G. Kramer A. Sigler A. Gosman A. Telemedicine as a means of effective speech evaluation for patients with cleft palate Ann Plast Surg 68 2012 415 417 22421491 22 Shenai M.B. Tubbs R.S. Guthrie B.L. Cohen-Gadol A.A. Virtual interactive presence for real-time, long-distance surgical collaboration during complex microsurgical procedures J Neurosurg 121 2014 277 284 24905563 23 Eberlin K.R. Perdikis G. Damitz L. Krochmal D.J. Kalliainen L.K. Bonawitz S.C. Electronic communication in plastic surgery: guiding principles from the American Society of Plastic Surgeons Health Policy Committee Plast Reconstr Surg 141 2018 500 505 29370003 24 Hamilton A.B. Brunner J. Cain C. Chuang E. Luger T.M. Canelo I. Engaging multilevel stakeholders in an implementation trial of evidence-based quality improvement in VA women’s health primary care Behav Med Pract Policy Res 7 2017 478 485 25 Glaser B.G. The constant comparative method of qualitative analysis Social Problems 12 1965 436 445 26 Benziger C.P. Huffman M.D. Sweis R.N. Stone N.J. The telehealth ten: a guide for a patient-assisted virtual physical examination Am J Med 134 2020 48 51 32687813 27 Douglas S. Geiger E. McGregor A. Norwich A. Abbate D. Hsia H. Telehealth in plastic surgery: a veterans affairs hospital perspective Plast Reconstr Surg Glob Open 6 2018 e1840 30534478 28 Yahanda A.T. Lafaro K.J. Spolverato G. Pawlik T.M. A systematic review of the factors that patients use to choose their surgeon World J Surg 40 2016 45 55 26362880 29 McLafferty R.B. Williams R.G. Lambert A.D. Dunnington G.L. Surgeon communication behaviors that lead patients to not recommend the surgeon to family members or friends: analysis and impact Surgery 140 2006 616 622 discussion 622–624 17011909 30 Weiner M. Biondich P. The influence of information technology on patient-physician relationships J Gen Intern Med 21 Suppl 1 2006 S35 S39 31 Axelrod D.A. Goold S.D. Maintaining trust in the surgeon-patient relationship: challenges for the new millennium Arch Surg 135 2000 55 61 10636348 32 Ansary AM, Martinez JN, Scott JD. The virtual physical exam in the 21st century. J Telemed Telecare. 2019; https://doi.org/10.1177/1357633X19878330. Accessed November 6, 2019. 33 Wiechert K. Wang J.C. Chapman J.R. Three pillars of expectation management in spine surgery: trust, communication, and patient education Global Spine J 9 2019 573 574 31448188 34 Hawley S.T. Hofer T.P. Janz N.K. Fagerlin A. Schwartz K. Liu L. Correlates of between-surgeon variation in breast cancer treatments Med Care 44 2006 609 616 16799355 35 Rahimi I. Gandomi A.H. A comprehensive review and analysis of operating room and surgery scheduling Arch Computat Methods Eng 2020 1 22 10.1007/s11831-020-09432-2 Epub ahead of print

33941389

PMC9754079

NO-CC CODE

2022-12-16 23:26:26

no

Surgery. 2021 Aug 1; 170(2):587-595

utf-8

Surgery

10.1016/j.surg.2021.03.029

oa_other

==== Front Sci Total Environ Sci Total Environ The Science of the Total Environment 0048-9697 1879-1026 Elsevier B.V. S0048-9697(22)00516-2 10.1016/j.scitotenv.2022.153424 153424 Article The case of “public congregation vs. COVID-19 PPE pollution”: Evidence, lessons, and recommendations from the annual pilgrimage to the Catholic Holy Site in Mexico City, Mexico Kutralam-Muniasamy Gurusamy a⁎ Shruti V.C. b⁎ a Department of Biotechnology and Bioengineering, Centro de Investigación y de Estudios Avanzados del Instituto Politécnico Nacional, Ciudad de México, Mexico b Instituto de Geología, Universidad Nacional Autónoma de México (UNAM), Ciudad Universitaria, Del. Coyoacán, C.P. 04510 Ciudad de México, Mexico ⁎ Corresponding authors. 25 1 2022 15 5 2022 25 1 2022 821 153424153424 24 12 2021 21 1 2022 22 1 2022 © 2022 Elsevier B.V. All rights reserved. 2022 Elsevier B.V. Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. Pollution from personal protective equipment (PPE), particularly face masks, has surfaced in the marine and terrestrial environments globally since the COVID-19 outbreak due to improper disposal practices and inadequate waste management, raising widespread alarm and attention. Our understanding of the prevalence and distribution of PPE in highly populated metropolitan areas is still emerging, and studies focusing specifically on developing countries in Latin America remain sparse. This study attempted to “kill two birds with one stone” by (1) addressing this knowledge gap by analyzing the degree of improper dispensing of PPE in Mexico City (Mexico) and (2) investigating the impact of massive public congregations on PPE contamination during the yearly pilgrimage to the Villa de Guadalupe on December 12th. Our survey findings revealed 731 PPE items within a 6-kilometer radius between December 5 and December 12, 2021, with daily densities ranging from 4.1 × 10−3–13.9 × 10−3 PPE items m−2. Face masks were the most disposed type of PPE (94%), with gloves and face shields accounting for just 6% of the total. The PPE disposal more than doubled as the pilgrim day approached, with an estimated disposal rate ranging from 151.52 to 506.06 items day−1, substantiating the surge in the disposal of used PPE to large public congregations that filled the surroundings during the pilgrimage. The observed average PPE density of 7.8 × 10−3 items m−2 was higher than in the metropolitan environments of Canada, Ghana, and Turkey. To our knowledge, this first study describes information showing the need to pay attention to the major impact of public events and mobility on COVID-19 PPE pollution, as well as emphasizes the necessity for adequate management facilities in improving PPE disposal. Graphical abstract Unlabelled Image Keywords COVID-19 Plastic pollution Environment Personal protective equipment Microplastics Waste management Editor: Damià Barceló ==== Body pmc1 Introduction Since the emergence of coronavirus disease (COVID) in China in late 2019, the world has been dealing with the COVID-19 pandemic, which has become the global health crisis of our time. Using a range of personal protective equipment (PPE), including face masks, wet wipes, gloves, shields, and aprons during the pandemic to limit the transmission of COVID-19 is one of the most effective human-made measures. Face masks of two types: single-use-disposable (melt-blown nonwoven fabrics and include N95 and surgical masks) and reusable (cloth masks) have been prevalent in recent years, which includes a variety of forms, colors, and designs with different filtering performance. Furthermore, polyurethane, polycarbonate, polypropylene, polystyrene, polyacrylonitrile, polyethylene, or polyester make up a large component of face masks (Fadare and Okoffo, 2020; Aragaw, 2020). The unprecedented demand, production, and usage of PPE, particularly face masks, has had a large impact on the plastic waste associated with COVID-19 PPE (Adyel, 2020). For example, Chowdhury et al. (2021) estimated that 0.15 million to 0.39 million tons of plastic waste could eventually end up in the coastal and marine environment. Poor PPE management and disposal practices can result in littering the terrestrial and marine habitats, and like other discarded plastic, they degrade slowly and remain in the environment for an extended period of time, causing widespread concern among the public and scientific community (Kutralam-Muniasamy et al., 2022). Furthermore, face masks have become a possible source of billions of microplastics after being discarded into the streets and into oceans (e.g., Morgana et al., 2021; Wang et al., 2021; Saliu et al., 2021; Ma et al., 2021; Li et al., 2021), releasing a variety of organic and inorganic pollutants into the environment (e.g., Liu and Mabury, 2021; Fernández-Arribas et al., 2021; De-la-Torre et al., 2022). Also, research has developed indicating that PPE, like other marine litter, can interact with biota. Ingesting PPE or the microplastics released by PPE could harm organisms´ health and end up in the food chain of aquatic and terrestrial environments (Kutralam-Muniasamy et al., 2022; Silva et al., 2021; Kwak and An, 2021). Thus, it is critical to comprehend where they are deposited and accumulated, as well as when the general public is exposed to discarded PPE, particularly face masks. This prompted the researchers to assess novel PPE pollution driven by the COVID-19 pandemic and identify the prevalence and distribution of disposed PPE items in a multitude of environments. Currently, PPE pollution has been well-documented in several oceans, tourist beaches (e.g., Thiel et al., 2021; De-la-Torre et al., 2021b, De-la-Torre et al., 2022; Akhbarizadeh et al., 2021; Rakib et al., 2021), and freshwater systems (e.g., Cordova et al., 2021), with the majority of investigations focusing on the marine ecosystem (Kutralam-Muniasamy et al., 2022). In view of the decrease in current SARS-CoV-2 instances, almost all countries have withdrawn or postponed their lockdown bans and have resumed activities under strict restrictions. Resuming activities will result in increased waste management burdens, and it is vital to understand how this process of normalization will unfold in terms of waste management and COVID-19 PPE pollution. Currently, only a few have investigated improperly discarded PPE in the streets of metropolitan cities near schools, hospitals, and residential areas in Canada (Ammendolia et al., 2021; France, 2021), Kenya (Okuku et al., 2021), South Africa (Ryan et al., 2020), Ghana (Amuah et al., 2021), Bangladesh (Abedin et al., 2022), and Turkey (Akarsu et al., 2021). These study findings strongly suggest that public mobility has a substantial effect on the disposal and buildup of PPE waste in metropolitan areas. Nonetheless, there is a dearth of knowledge on how public mobility associated with massive social gatherings impacts COVID-19 PPE pollution in densely populated areas, and studies on this issue are lacking. We sought to draw attention to massive public gatherings since they are well recognized for their potential to generate hundreds of tons of solid waste in a short period of time, posing challenges to local governments and necessitating extra-care specialized waste management solutions, particularly in developing countries. In current work, we aim to investigate the changing patterns of occurrence, characteristics, and density of PPE items during the yearly pilgrimage of millions of people to the Villa de Guadalupe in Mexico City (Mexico), in order to better understand the impacts of massive public gatherings on PPE littering during the COVID-19 pandemic. By doing so, this study achieved “two birds with one stone” by: (1) addressing the knowledge gap of COVID-19 PPE pollution in Latin American metropolitan areas and (2) investigating the influence of massive public gatherings on PPE pollution in the urban environment. Therefore, the findings of this study will help to identify societal behavioral trends regarding PPE improper disposal during public gatherings, and this data will be crucial in laying the groundwork for future waste management tools and regulations in Mexico City. It also serves as baseline information for the general public, researchers, the media, and government authorities in order for them to adopt and enhance efforts to prevent PPE contamination at future public gatherings. 2 Methodology 2.1 Study area Mexico City is one of the world's largest and most densely populated metropolitan regions, housing nearly 21 million inhabitants. In the lively Mexico City neighborhood of Tepeyac, stands the most-visited religious site in the West: The Villa de Guadalupe (Basilica of Guadalupe). This national shrine receives as many as twenty million pilgrims annually from villages (pueblos), cities, and suburbs, and even across international borders. Every year in mid-December, millions of pilgrims surge up a broad avenue in Mexico City toward the Villa de Guadalupe from the states around Mexico City, including Tlaxcala, Puebla, Estado de México, and Querétaro for the “Día de la Virgen.” In Mexico, this religious pilgrimage has been practiced for generations. For example, on December 12th, 2019, more than 10 million pilgrims from various Mexican states visited the Basilica de Guadalupe (GDF, 2019). Due to the COVID19 pandemic circumstances, there were no pilgrimage activities in 2020. With the resumption of activities, a larger number of visitors are expected on this yearly pilgrimage to the Villa de Guadalupe. With this in mind, the state government has taken a number of immediate actions. Only those having a complete COVID-19 vaccine system were encouraged to participate in the pilgrimage activities, and masks were required as well as supplied if necessary. Pilgrims were not permitted to remain within the atrium or temples. The local government has placed more than a hundred sanitation workers and volunteers to manage the solid waste resulting from the pilgrimage. More importantly, the government has requested people schedule their visits before December 12th to avoid social overcrowding and contact among people. Taking into account the constraints and understanding the prevailing COVID-19 pandemic, pilgrims from the surrounding states began making the pilgrimage earlier on December 12th (Azteca Noticias, 2021). Thus, the study was conducted between December 5 and December 12, 2021, to evaluate the improperly disposed PPE items. Moreover, we have limited our PPE littering survey until December 12 since public camping places will close after that day, and most pilgrims will not go beyond that date. 2.2 Sampling and PPE analysis We monitored PPE littering at three different locations within a 6-kilometer radius of the Villa de Guadalupe, as shown in Fig. 1 . The three sites include camping sites 1 and 2, and a walking site. Camping sites 1 and 2 were mainly selected due to the presence of intensive human activities during the pilgrimage season. Camping Site 1 is a recreational park (Parque del Mestizaje) where pilgrims have a short-stay until they complete their rites, while Camping Site 2, often known as “casa de peregrinos” or “house of pilgrims,” is a government-run facility that is only open to pilgrims during the month of December. The distance walked by pilgrims between camping sites 1 and 2 and the Villa de Guadalupe is referred to as the “walking site.” Furthermore, the sampling sites cover a cumulative area of approximately 36,536 m2 (camping site 1: 3446 m2; walking site: 14,054 m2; camping site 2: 19,036 m2). To avoid any health risks associated with the pandemic, rigorous safety protocols were taken during data collection, always including the use of masks, social distancing, and hand sanitizer. In addition, local restrictions were followed as advised during the pilgrimage. We began monitoring PPE disposal every day at 10:00 a.m. local time and continued for 6 h before scheduled sanitation personnel cleaning operations in order to avoid bias. This time interval for PPE monitoring was chosen due to the increasing number of pilgrims to the Villa de Guadalupe. We walked a 6 km radius to look for PPE discarded on the streets, under cars, in parking lots, and near camping areas. To ensure that the disposed PPE items were covered thoroughly, one person went along one side of the roadway while the other walked down the other. Dumpsites throughout the camping area were also inspected to ensure the accuracy of the PPE disposal data. Visually identified PPEs were photographed and counted.Fig. 1 Map of the study area, Villa de Guadalupe (Mexico City), and sample locations. Fig. 1 The density of PPE items was calculated using the following equation (Okuku et al., 2021):C=n/A where C is the density of PPE items per m2, n is the number of PPE, and A is the surveyed area (m2). Daily PPE release in this study was estimated using the following equation:D=n/t where D is the daily release of PPE (items/day), n is the number of collected undamaged PPE items on the second day of sampling, and t is observation time (day). 2.3 Statistical analysis The obtained data were grouped by sites to determine their influence on the PPE density. The Shapiro-Wilk test invalidated the data's normal distribution (p < 0.05); hence, non-parametric tests were performed. Kruskal-Wallis and Dunn's multiple comparison tests were employed to analyze and compare whether there were any significant differences between sites. The statistical significance threshold of 0.05 was used. GraphPad Prism (version 7 for Windows) was used to carry out all statistical analyses. 3 Results and discussions PPE litter was recorded in all three sampling locations and on all days throughout the study period. Face masks, face shields, and gloves were among the PPE items visually identified. For wipes, we adopted the same method as Ammendolia et al. (2021) to distinguish paper-based tissues or paper towels from synthetic wipes. We found that the disposed wipes were paper-based rather than synthetic, and as a result, there were no synthetic wipes in our surveys. A total number of 731 PPE items was found over a cumulative area of 36,536 m2 for the three sample locations. Throughout the survey, improperly disposed PPE materials were visually identified, counted, and photographed in a multitude of urban settings, as seen in Fig. 2 . Face masks accounted for 94% of the total (n = 689), with other PPE items (i.e., gloves and face shields) accounting for the remaining 6% (n = 42) (Fig. 3 ). We classified the face masks based on their types and colors. A variety of colored (black, blue, yellow, purple, green, and pink) and white mask items from surgical, KN95, and cloth were identified. There were face masks for adults and children in the discarded PPE items, with adult face masks accounting for 99% of the total. The face masks that were spotted were either strapless or intact, but not torn or damaged. The surgical masks documented were mostly double-layered, with no three-layered masks with disposable layers found. KN95 masks with or without respirators were also observed. As shown in Fig. 3, surgical masks dominated with 87% of the total (n = 596), followed by KN95 (n = 56; 8%) and reusable cloth masks (n = 37; 5%). When the face masks were sorted by color, it was determined that black and blue-colored masks made up 84% (n = 578) of the total masks. Other face masks that composed 17% (n = 111) of the total were white (n = 42; 6%), multicolored/designed (n = 26; 4%), pink (n = 21; 3%), green (n = 9; 1%), purple (n = 10; 1.5%), and yellow (n = 3; 0.5%) (Fig. 3).Fig. 2 A few examples of different colors and types of face masks spotted at the sampling locations from the streets, parking lot, under the vehicle, and near the sewer system. Fig. 2 Fig. 3 Quantity, distribution, and characteristics of disposed PPE items in the vicinity of the Villa de Guadalupe (Mexico City). (a) PPE abundance by day and sampling sites. (b) Percentage composition of littered face mask items for the whole study period. Others: gloves and face shields. Fig. 3 The relative abundance and density of PPE items recorded in the study area are shown in Table 1 . Fig. 3 shows the distribution of PPE items from the sampling sites. The abundance of disposed PPE items during our survey were in the range of 50 to 167, with an average disposal of 91.375 PPE items. The number of PPE items was significantly higher on the first sampling day than on subsequent sampling days (Days 2–5), indicating that pilgrims arrived earlier to perform religious rituals throughout the weekend and left PPE items in the areas where they camped and walked. Furthermore, no significant variations were identified from sample days 2 to 5, indicating that the number of pilgrims was lower compared to the first day of sampling and that locals had a bigger effect on PPE littering. Nonetheless, the abundance of PPE items in the next three consecutive sampling days (Days 6–8) significantly increased as the pilgrimage day approached. The abundance for sample days 6, 7, and 8 was 119, 142, and 167 PPE items, respectively. It should be noted that the abundance from the final three days of sampling was 428 PPE items, which is equivalent to 58% of the total PPE items detected in this study. According to government officials, the number of pilgrims visiting the Villa de Guadalupe between December 10 and December 12, 2021 (sample days 6 to 8) has grown massively (GDF, 2021). This dramatic increase in the number of pilgrims to the Villa de Guadalupe substantiated the surge in the disposal of PPE items in the sample days 6 to 8. At the same time, it is worth noting that there were no improperly discarded PPE items within the atrium or temple for the entire study, reflecting pilgrims' and the general public's responsible behavior. However, given the Villa de Guadalupe's surroundings, we cannot say the same because of the prevalence of improperly disposed PPE items within a 6 km radius across the study period.Table 1 Relative abundance of PPE items during the yearly pilgrimage to Villa de Guadalupe 2021. Table 1Date Total (PPE items h−6) Average (PPE items h−6) Daily release (items day−1) Density (PPE items m−2) 12/05/21 75 25 7.8 × 10−3 12/06/21 50 16.67 151.52 4.1 × 10−3 12/07/21 56 18.67 169.70 4.6 × 10−3 12/08/21 64 21.33 193.94 5.3 × 10−3 12/09/21 58 19.33 175.76 4.8 × 10−3 12/10/21 119 39.67 360.61 10 × 10−3 12/11/21 142 47.33 430.30 11.8 × 10−3 12/12/21 167 55.67 506.06 13.9 × 10−3 It is unfortunate that some negative pandemic-related behaviors of the public with regard to PPE disposal tend to persist in metropolitan areas. This might be due to a lack of public awareness and social responsibility regarding PPE disposal; in addition, they may be unaware that they are dealing with infectious waste, either knowingly or unknowingly, whose disposal could have serious environmental and health concerns. Because the face masks were single-use disposable, the pilgrims likely did not find them useful when returning from their pilgrimage, necessitating their disposal. During our PPE evaluation, we noticed that the local government has installed numerous organic and inorganic containers for depositing solid waste, taking into account past years' solid waste experiences. However, because PPE waste was unfamiliar to everyone prior to the COVID-19 pandemic and there was no previous understanding of this waste, only a limited number of PPE disposal bins have been established to drop the used face masks and gloves into. More importantly, the PPE disposal bins have not been evenly distributed considering the human hotspots during the pilgrimage. Similarly, there were no signs suggesting that PPE items should not be discarded on the streets but should instead be disposed of in the bins located around the Villa de Guadalupe. Hence, one of the primary reasons for increased littering and improper disposal in the streets and surrounding areas can be attributed to a lack of adequate PPE disposal bins. Similarly, Ammendolia et al. (2021) observed similar limitations in Canada's Toronto, attributable to inappropriate disposal of used face masks in metropolitan settings. The disposed PPE items were spotted on the streets, near the atrium, underneath cars, trucks, and bikes, in the parking lot, and near sewer systems (Fig. 2). As seen in Fig. 2, a few face masks were deliberately thrown down, owing to straps breaking or loosening while others were being spun about and dumped. And, at times, the masks spotted under vehicles and in the parking lots appeared to have been quickly discarded, leaving the locations for no apparent reason. Children's masks would have fallen to the ground and been abandoned instead of being properly disposed of in the bins while they were playing or wandering around. Among the abandoned PPE items, black and blue surgical masks predominated because they were less expensive and more widely accessible than KN95 during the pilgrimage. In addition, the local state government had taken the necessary precautions and had the authority to distribute face masks, particularly blue masks, to those in need. When pilgrims were mandated to wear face masks, local shopkeepers grabbed the opportunity, and the number of stores selling masks items (particularly blue and black surgical masks) within a 6 km radius more than quadrupled, contributing to unprecedented supply, use, and PPE littering. There is still a fundamental point to be asked: Are public gatherings the critical players driving the disposal of PPE items in the Villa de Guadalupe's surroundings? We answer yes, and more evidence can be obtained by analyzing the changing pattern of the number of PPE items based on the locations studied (Fig. 4 ). There was a significant difference in the amount of PPE items between the camping sites (CS1 and CS2 combined) and walking site. Fig. 4 shows the PPE littering was found higher in the camping sites than in the walking site. The number of PPE items ranged from 19 to 112 in the camping sites and from 24 to 55 in the walking sites. Camping sites accounted for two-thirds (n = 458) of the total, with walking sites accounting for only 37% (n = 273). As shown in Fig. 4, the number of discarded PPE items remained relatively consistent with no significant differences throughout sample days 1–5 and sampling sites. However, a greater number of PPE items were found on next sample days 6 to 8 in the camping sites (n = 286) than walking site (n = 141). The dominance, with a two-fold rise in disposed PPE items at the camping site, could be attributed to a greater number of pilgrims from outside the city during the pilgrimage in the study area. It is essential to note that the locations of the camping sites are the only places where pilgrims can stay for a brief period of time during their annual pilgrimage until they return. And any discarded PPE items detected in these areas must be ascribed to pilgrims rather than locals living nearby to the surroundings. The observed rise in the concentration of PPE littered at the camping site as the pilgrims approached on sampling days 6 to 8 is clearly indicative of the increased disposal behavior of pilgrims staying and their influence on the discarded items in the study area. It shows that the surroundings of the Villa de Guadalupe in Mexico City suffer from PPE littering during the pilgrimage due to the contributions of the pilgrim population, and there has been less guidance from public sanitation officials to pilgrims on how to dispose of PPE items for environmental and health protection. Furthermore, the data evidence leads to two major findings: (1) Public mobility within cities and neighborhoods during a major event (say national or regional) roots for PPE littering and contamination. (2) Their short stays in designated areas will not only contribute to solid waste, but will also increase the amount of PPE littered in the surrounding area, posing additional challenges and barriers to waste management.Fig. 4 The distribution of COVID-19-related PPE litter by sample site is heterogeneous. Color (a) is at the top, and type (b) is at the bottom. Fig. 4 According to local government officials, the total number of pilgrims in 2021 was found to be 63% lower than in 2019 (GDF, 2021). We believe that if the projected pilgrims, such as those in 2019, gathered in Mexico City, the amount of PPE items disposed of improperly would be more than the number seen in this study. Furthermore, the information on the PPE littering the pilgrimage route is unclear because the bulk of pilgrims have traveled 120–250 km to Mexico City in autos and on foot, necessitating attention into this subject. At the same time, based on our findings, we found that the PPE density ranged from 4.1 × 10−3 to 13.9 × 10−3 items day−1, with an average density of 7.78 × 10−3 items day−1. As expected, the average density of PPE items encountered greater in camping site (1.2 × 10−3 PPE items day−1) than walking site (1.03 × 10−3 PPE items day−1). The Kruskal-Wallis test revealed significant differences (Chisquare = 10.39, p = 0.0022) in PPE density between the three sites. According to Dunn's multiple comparison test, the PPE density at camping site 1 differed substantially from camping site 2 (p = 0.0042), but not from walking site (p = 0.1416), as shown in Fig. 5 . Moreover, the estimated daily release of PPE items was between 151.52 and 506.06 items day−1, with an average release of 283.98 items day−1.Fig. 5 Boxplot of the PPE density grouped per the camping site 1 (CS1), walking site (WS), and camping site 2 (CS2) using Kruskal-Wallis with Dunn's multiple comparison test. Equal letters indicate no significant differences, while different letters indicate significant differences. Fig. 5 Will the discarded PPE items be a source of pollution and a threat to the environment? PPE pollution in metropolitan areas may not be as impacting as in coastal or natural areas. However, once discarded, PPE litter in the metropolitan area has a variety of fates that have an impact on the environment, either directly or indirectly. In one scenario, they might be collected, sorted in a biomedical bag, and incinerated with biomedical waste, adding to CO2 emissions into the atmosphere and impacting global warming. Another possibility is that they will be mixed with municipal solid waste and dumped in landfills. Nonetheless, dumping PPE waste in landfills without adequate treatment has the potential to increase worldwide plastic pollution and the spread of SARS-CoV-2. In addition, PPE waste degrades into microplastics in anaerobic conditions through a variety of physical and chemical processes, which are ultimately deposited in the living environment (Shruti et al., 2020). On the one hand, the long-term deposition of the microplastics in the soil can decrease soil fertility and hamper plants growth. On the other hand, they may leach out of the soil and into neighboring water ways, where they could be consumed by organisms. Given the possibility of a shortage of sanitation personnel, which is especially troublesome in low-resource countries, PPE items may have been inadequately and infrequently removed. Improper collection of PPE wastes would most likely endanger the health of the general public and healthcare personnel. PPE waste negatively influences the livelihoods of the surrounding community as well as visitors to such valuable locations. If these abandoned PPE items are not properly removed, children playing nearby may come into contact with them, causing risk of disease transfer and hazards associated to it. Animals (e.g., stray dogs) and birds (e.g., crows) in cities are more likely to interact and torn the PPE items, either because they mistake them for food or because they are seeking for food. These PPE items are sometimes released directly into aquatic ecosystems or by stormwater and rain runoff that ends up in the sewage drain and finally, wastewater treatment plants. Face masks found alongside the drainage system, as depicted in Fig. 2, can enter the sewer drain immediately due to wind, human activity, or rainwater runoff. Face masks are likely to release a substantial proportion of microfibers comprised of petrochemical polymers such as PP and PE (e.g., Morgana et al., 2021; Wang et al., 2021; Saliu et al., 2021; Ma et al., 2021; Li et al., 2021), as well as their associated organic (e.g., phthalates, antioxidants, organophosphate esters, bisphenols, and plastic additives) (e.g., Liu and Mabury, 2021; Fernández-Arribas et al., 2021) and inorganic (metals such as Zn, Mn, Ti, Fe, and Ca) (e.g., De-la-Torre et al., 2022) pollutants when they reach the aquatic environment or a landfill. Otherwise, the discarded face masks in areas (such as the wilderness in the park) inaccessible to sanitation staff will be exposed to UV light, which will degrade and release microplastics and smaller plastic items in their surroundings until they are collected. In any instance, PPE waste is prone to degradation, releasing micropollutants into the environment. These micropollutants from face masks have widely been demonstrated to have harmful effects on the environment and human health (Kutralam-Muniasamy et al., 2022; Silva et al., 2021). Previously, a few studies validated the occurrence and distribution of improperly disposed PPE items in urban areas (Table 2 ). Some have examined and reported on PPE disposal findings for weekdays and weekends (e.g., Amuah et al., 2021), while others have done so for a specific time period (e.g., Ammendolia et al., 2021). However, the PPE waste improperly disposed of as a result of responses to public mobility has not been examined or explored in relation to large gatherings in metropolitan areas. To the best of our knowledge, this is the first study that has examined and presented evidence on the subject. Our findings indicate a large number of incidents of PPE improper disposal can be expected in public congregations in metropolitan locations. This is congruent with the plethora of information gained in marine environments, which has linked the rising number of beach visitors to higher PPE disposal (Thiel et al., 2021; De-la-Torre et al., 2021; Hassan et al., 2021). According to the previous studies, the number of face mask items (n = 578) observed in this study was higher than in Turkey (n = 546), Ghana (n = 535), and Canada (n = 274) (Table 2). The observed average PPE density of 7.8 × 10−3 items m−2 was higher than in the metropolitan environments of Canada (1.01 × 10−3 ± 1.55 × 10−3 items m−2 and 0.0001 ± 0.00005 items m−2). The probable reasons for the differences in the PPE density in metropolitan areas reported in the literature are highly dependent on (i) the type of sampling method used, (ii) the area covered, (iii) the length of sampling, (iv) lockdown restrictions, and (v) public mobility. Despite the lack of a standardized approach for assessing PPE disposal, studies have used the same unit and reported the PPE density in terms of items per m−2. Regardless of the differences in the number of PPE items found, surgical face masks were more abundant in previous investigations, as was the case in our study. While we only observed 3% of cloth masks, France (2022) found 23% of cloth masks, showing that surgical masks are preferred over reusable masks in this region of the world. In contrast to previous investigations, no synthetic wet wipes were found in any of our surveys. Despite significant efforts to control and manage PPE waste, available research indicates that it is prevalent in the urban environment. Furthermore, the degree of PPE contamination in urban environments is largely unknown all across the world, necessitating future investigation. It is worth keeping in mind that only by monitoring will we be able to gain information on how current waste management strategies have performed and how to improve them.Table 2 Recent studies on littered PPE in metropolitan locations worldwide. Table 2Study location Environment Area surveyed (m2) and number of days Sampling location Methods Key findings Reference Kenya City NR; 30 Streets 1) A distance between 200 m and 2000 m was surveyed 2) PPE litter was picked from a width of 2 m on both sides of the street 1) 0–5.6 × 10−2 PPE m−2 Okuku et al., 2021 South Africa Urban NR; 50 Streets 1) Litter was collected from 400 m of street margins 1) Face masks and gloves contributed <1% of total mass 2) Found wet wipes amid other PPE Ryan et al., 2020 Ghana City NR Streets 1) Survey by transects covering a distance of 100 m to 200 m of different areas like township, suburb, institution, municipal and community 2) Visually identified, counted and photographed 1) Total of 535 face masks along 1720 m stretch 2) Density range 0.04 m to 0.42 m Amuah et al., 2021 Canada City 245,190; 34 Street, under cars, residential areas, grocery, and hospital zones 1) Collected debris that was 1 m and 5 m from the closest edge of the sidewalk 1) Face masks constituted 31% (n = 274) of total plastic debris 2) 95% of disposable face masks, 3% reusable masks, and only 2 high-grade masks such as N95 and KN95 3) Disinfecting wipes constituted 25% of total plastic debris 4) 0–8.22 × 10−3 PPE m−2; mean density: 1.01 × 10−3 ± 1.55 × 10−3 PPE m−2 Ammendolia et al., 2021 Canada City 750,000; 10 Streets, Highway Montreal-Quebec 1) 3 m wide ground surface survey by walking covering urban, town, and rural areas 1) 0.0001 ± 0.00005 items m−2; 2) Per day total range 4 to 10 masks 3) 76% Surgical masks and 24% cloth masks France, 2021 Turkey City 40,000; NR Streets 1) A perimeter of 1 km2 in three cities was surveyed and the masks found within these perimeters were collected. 1) 30 to 96 face masks km2 2) Total 210 for Adana, 170 for Mersin, and 166 for Niğde region Akarsu et al., 2021 Mexico City 36,536; 8 Streets, under cars, and camping areas 1) Surveyed 6 km radius of the pilgrimage site. 2) PPE litter was monitored along both sides of the road and pilgrimage camping sites. 3) Visually identified, counted, and photographed 1) Abundance: 4.1 × 10−3–13.9 × 10−3 items m−2; mean density: 7.8 × 10−3 items m−2 2) White and colored (black, blue, pink, yellow and green) face masks were observed 3) 81% surgical face masks, 8% KN95, 5% cloth and 6% others (gloves, face shields) This study NR: Not reported; PPE: Personal Protective Equipment. 4 Recommendations for tackling a new challenge: COVID-19 PPE littering in public gatherings Waste management measures are still vital at this point in the pandemic. The handling of PPE waste calls for immediate attention and the adoption of the best practicable approach toward preserving the environment and protecting human health. It is recognized that the effective management of plastic waste in developing countries with densely populated areas has been a great challenge even before the onset of the COVID pandemic. And recently, with the increased use and disposal of PPE, it has worsened the concerns about plastic waste management due to the safety challenges in handling infectious PPE waste. The CDMX government reported that about 510.2 t of waste were retrieved during this year's annual pilgrimage. We appreciate the government's efficient waste management, which includes numerous sanitation staff, three cleaning shifts in and around the Villa de Guadalupe, free mask distribution, medical facilities, and security. Several studies have addressed the problem of PPE waste and management strategies for use in diverse environmental situations such as coastal cities, beaches, and metropolitan areas. Thiel et al. (2021), for example, suggested installing sufficient and strategically placed waste-bins for discarded PPE materials and enforcing strict safety regulations for people in beach environments. Similarly, Ammendolia et al. (2021) called for the installation of numerous PPE waste bins in Toronto (Canada) metropolitan areas to encourage residents to properly dispose of PPE items, as well as the use of reusable PPE items (e.g., cloth masks and cloth gloves) in place of plastic-based PPE. However, our understanding of strategies for managing PPE waste during massive public events in metropolitan areas remains in its infancy. Given the potential environmental and human health risks posed by intentionally or unintentionally littering PPE in large public gatherings, there is an urgent need for a comprehensive monitoring system for PPE disposal and to identify hotspot areas that allow authorities to implement timely and effective management strategies tailored to large public gatherings. Nonetheless, implementing waste management measures during a pandemic is vastly different and far more challenging than under normal conditions for large public gatherings. It mainly depends on the ways in which societies respond. To assist the government and other organizations in developing efficient PPE waste management programs for large public events, we suggest the following, as illustrated in Fig. 6 : (1) Raising funds from the government and local authorities to carry out effective PPE waste management during the event; (2) Identifying and placing waste bins and sanitation staff in public gathering hotspots (e.g., camp sites) with the assistance of locals; (3) Installing waste bins exclusively for PPE disposal with hand sanitizer for every 100 f. of distance and ensuring equitable access and distribution throughout the event area; and (4) Providing free face masks to attendees; (5) Distributing flyers with clear instructions on safety precautions and proper disposal of PPE items throughout the event; (6) Placing sign boards with clear instructions on proper disposal of PPE items; (7) Frequent announcements in regard to safety measures, social distancing, and correct disposal of PPE items; (8) Disinfecting the public before they enter the event area either by installing a disinfection tunnel or by using disinfectant spray; (9) Providing access to people only after a thorough body temperature check, a vaccination record, and the necessary use of a mask; (10) Promoting the use of reusable masks and gloves; recruiting sanitation personnel and NGO's; (11) Separately collecting and transporting used masks, gloves, personal clothing, and all PPE in closed medical waste bags to final treatments (e.g., landfilling or incineration); and (12) Provide the event's official announcement with the safety measurements and PPE disposal instructions. Last but not least, simply managing PPE waste is insufficient. After managing, there should be a sort of sustainable valorization options (for example waste to energy). The conversion of PPE to oil- and bio-fuels via pyrolysis is one potential recycling solution for the huge amount of synthetic polymer-based PPE generated during the pandemic (Aragaw and Mekonnen, 2021).Fig. 6 Recommendations for addressing new COVID-19 PPE pollution in the case of large - scale public gatherings. Fig. 6 5 Conclusion The COVID-19 pandemic has continued to have a major impact on global plastic pollution. Recent studies have substantially advanced our understanding of PPE contamination in the environment during the COVID-19 pandemic. This study investigated and documented evidence of improper PPE waste disposal and management during public gatherings, as well as its contribution to increased PPE littering in the Villa de Guadalupe neighborhood of Mexico City. To our knowledge, this is the first attempt to raise concerns and provide answers about whether large public events may contribute to PPE pollution in metropolitan areas, thus contributing valuable information to ongoing efforts on the topic of PPE pollution and improving current waste management practices. According to our findings, PPE littering, particularly of face masks, is prevalent across city streets, parking lots, atriums, and recreational areas, increasing environmental plastic pollution in densely urbanized areas. Also, the areas open to the public for camping and short stays during pilgrimage have had higher improper dispensing of PPE waste. These identified sources of information and new data can be used to build solutions to aid waste management during future events or public gatherings. Our findings support previous researchers' views on the importance of proper waste management, and increasing public awareness will be critical to achieving effective PPE waste management. On the one hand, metropolitan government officials must conduct educational campaigns to raise awareness regarding PPE disposal and pollution, as well as take long-term measures for handling PPE waste. Researchers, on the other hand, could undertake citizen science programs as part of their investigation process in order to widen environmental awareness regarding plastic pollution among the local population. With unforeseen outbreaks continuing to occur throughout the world, we believe that the COVID-19 pandemic will extend for several years, necessitating the use of PPE. More studies are warranted to assess the ongoing plastic pollution burden associated with discarded COVID-19 PPE wastes, as well as to uncover the factors that contribute to PPE pollution in the environment. As the literature on environmental PPE monitoring expands, standardized approaches for comparing and integrating data from worldwide PPE screens will be essential. Given the uprising of large public gatherings, we must take measures to limit the release of used PPE into the environment. Many concerts and entertainment events with massive public gatherings, for example, are generally recognized to take place in coastal regions, and beach concert series are resuming after a prolonged gap. If improper disposal from a large public gathering in a coastal or beach area is not effectively monitored and managed, it will provide a direct conduit of PPE items into the local aquatic ecosystems. The exercise in this study could be expanded with the appropriate design and implementation in order to better understand the impacts of beach events on PPE littering, which is currently unexplored and requires major attention. Furthermore, the knowledge gained from this study and the recommended strategies can be applied to similar areas where major public events are contemplated. CRediT authorship contribution statement Gurusamy Kutralam-Muniasamy: Conceptualization, Methodology, Data curation, Writing – original draft. V.C. Shruti: Conceptualization, Methodology, Data curation, Writing – original draft. Declaration of competing interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Acknowledgments VCS gratefully acknowledges financial support from DGAPA-UNAM postdoctoral fellowship program, Instituto de Geología, 10.13039/501100005739 Universidad Nacional Autónoma de México . Gurusamy Kutralam-Muniasamy and V.C. Shruti would like to express our gratitude to our mentor Dr. Fermín Pérez-Guevara (Biotechnology and Bioengineering, CINVESTAV) for his years of invaluable assistance in our scientific endeavors. This work is dedicated to our grandmother, Ambujam Rangarajan, who passed away on December 6, 2021. The authors would like to acknowledge all the doctors, health-care professionals, police personnel, sanitation workers and waste collectors at the frontlines working silently and tirelessly during COVID-19 outbreak world-wide. ==== Refs References Abedin M. Khandaker M.U. Uddin M. Karim M. Ahamad M. Islam M. Arif A.M. Sulieman A. Idris A.M. PPE pollution in the terrestrial and aquatic environment of the Chittagong city area associated with the COVID-19 pandemic and concomitant health implications Environ. Sci. Pollut. Res. 2022 1 13 Adyel T.M. Accumulation of plastic waste during COVID-19 Science 369 2020 1314 1315 32913095 Akarsu C. Madenli Ö. Deveci E.Ü. Characterization of littered face masks in the southeastern part of Turkey Environ. Sci. Pollut. Res. 2021 1 11 Akhbarizadeh R. Dobaradaran S. Nabipour I. Tangestani M. Abedi D. Javanfekr F. Jeddi F. Zendehboodi A. Abandoned Covid-19 personal protective equipment along the Bushehr shores, the Persian Gulf: an emerging source of secondary microplastics in coastlines Mar. Pollut. Bull. 168 2021 112386 Ammendolia J. Saturno J. Brooks A.L. Jacobs S. Jambeck J.R. An emerging source of plastic pollution: environmental presence of plastic personal protective equipment (PPE) debris related to COVID-19 in a metropolitan city Environ. Pollut. 269 2021 116160 Amuah E.E.Y. Agyemang E.P. Dankwa P. Fei-Baffoe B. Kazapoe R.W. Douti N.B. Are used face masks handled as infectious waste? Novel pollution driven by the COVID-19 pandemic Resour. Conserv. Recycl. Adv. 2021 200062 34939066 Aragaw T.A. Surgical face masks as a potential source for microplastic pollution in the COVID-19 scenario Mar. Pollut. Bull. 159 2020 111517 Aragaw T.A. Mekonnen B.A. Current plastics pollution threats due to COVID-19 and its possible mitigation techniques: a waste-to-energy conversion via pyrolysis Environ. Syst. Res. 10 1 2021 1 11 Azteca Noticias https://www.tvazteca.com/aztecanoticias/peregrinos-basilica-guadalupe-mtp 2021 Accessed Dec 2021 Chowdhury H. Chowdhury T. Sait S.M. Estimating marine plastic pollution from COVID-19 face masks in coastal regions Mar. Pollut. Bull. 168 2021 112419 Cordova M.R. Nurhati I.S. Riani E. Iswari M.Y. Unprecedented plastic-made personal protective equipment (PPE) debris in river outlets into Jakarta Bay during COVID-19 pandemic Chemosphere 268 2021 129360 De-la-Torre G.E. Rakib M.R.J. Pizarro-Ortega C.I. Dioses-Salinas D.C. Occurrence of personal protective equipment (PPE) associated with the COVID-19 pandemic along the coast of Lima,Peru Sci. Total Environ. 774 2021 145774 De-la-Torre G.E. Dioses-Salinas D.C. Pizarro-Ortega C.I. Fernández Severini M.D. López A.D.F. Mansilla R. Ayala F. Castillo L.M.J. Castillo-Paico E. Torres D.A. Mendoza-Castilla L.M. Meza-Chuquizuta C. Vizcarra J.K. Mejía M. De La Gala J.J.V. Ninaja E.A.S. Calisaya D.L.S. Flores-Miranda W.E. Rosillo J.L.E. Espinoza-Morriberón D. Gonzales K.N. Torres F.G. Rimondino G.N. Ben-Haddad M. Dobaradaran S. Aragaw T.A. Santillán L. Binational survey of personal protective equipment (PPE) pollution driven by the COVID-19 pandemic in coastal environments: abundance, distribution, and analytical characterization J. Hazard. Mater. 426 2022 128070 10.1016/j.jhazmat.2021.128070 Fadare O.O. Okoffo E.D. Covid-19 face masks: a potential source of microplastic fibers in the environment Sci. Total Environ. 737 2020 140279 Fernández-Arribas J. Moreno T. Bartrolí R. Eljarrat E. COVID-19 face masks: a new source of human and environmental exposure to organophosphate esters Environ. Int. 154 2021 106654 France R.L. First landscape-scale survey of the background level of COVID-19 face mask litter: exploring the potential for citizen science data collection during a ‘pollution pilgrimage’ of walking a 250-km roadside transect Sci. Total Environ. 2021 151569 34774631 GDF Gobierno de la Ciudad de México https://www.jefaturadegobierno.cdmx.gob.mx/comunicacion/nota/reporta-gobierno-de-la-ciudad-de-mexico-saldo-blanco-durante-el-operativo-basilica-2019 2019 Accessed on 22nd Dec, 2021 GDF Gobierno de la Ciudad de México https://jefaturadegobierno.cdmx.gob.mx/comunicacion/nota/visitaron-la-basilica-de-guadalupe-35-millones-de-peregrinos 2021 Accessed on 22nd Dec, 2021 Hassan I.A. Younis A. Al Ghamdi M.A. Almazroui M. Basahi J.M. El-Sheekh M.M. Abouelkhair E.K. Haiba N.S. Alhussaini M.S. Hajjar D. Wahab M.M.A. Contamination of the marine environment in Egypt and Saudi Arabia with personal protective equipment during COVID-19 pandemic: a short focus Sci. Total Environ. 2021 152046 34856280 Kutralam-Muniasamy G. Pérez-Guevara F. Shruti V.C. A critical synthesis of current peer-reviewed literature on the environmental and human health impacts of COVID-19 PPE litter: new findings and next steps J. Hazard. Mat. 422 2022 126945 Kwak J.I. An Y.J. Post COVID-19 pandemic: biofragmentation and soil ecotoxicological effects of microplastics derived from face masks J. Hazard. Mat. 416 2021 126169 Li L. Zhao X. Li Z. Song K. COVID-19: performance study of microplastic inhalation risk posed by wearing masks J. Hazard. Mater. 411 2021 124955 Liu R. Mabury S.A. Single-use face masks as a potential source of synthetic antioxidants to the environment Environ. Sci. Technol. Lett. 8 8 2021 651 655 Ma J. Chen F. Xu H. Jiang H. Liu J. Li P. Chen C.C. Pan K. Face masks as a source of nanoplastics and microplastics in the environment: quantification, characterization, and potential for bioaccumulation Environ. Pollut. 288 2021 117748 Morgana S. Casentini B. Amalfitano S. Uncovering the release of micro/nanoplastics from disposable face masks at times of COVID-19 J. Hazard. Mater. 419 2021 126507 Okuku E. Kiteresi L. Owato G. Otieno K. Mwalugha C. Mbuche M. Gwada B. Nelson A. Chepkemboi P. Achieng Q. Wanjeri V. The impacts of COVID-19 pandemic on marine litter pollution along the Kenyan Coast: a synthesis after 100 days following the first reported case in Kenya Mar. Pollut. Bull. 162 2021 111840 Rakib M.R.J. De-la-Torre G.E. Pizarro-Ortega C.I. Dioses-Salinas D.C. Al-Nahian S. Personal protective equipment (PPE) pollution driven by the COVID-19 pandemic in Cox's Bazar, the longest natural beach in the world Mar. Pollut. Bull. 169 2021 112497 Ryan P.G. Maclean K. Weideman E.A. The impact of the COVID-19 lockdown on urban street litter in South Africa Environ. Process. 7 4 2020 1303 1312 Saliu F. Veronelli M. Raguso C. Barana D. Galli P. Lasagni M. The release process of microfibers: from surgical face masks into the marine environment Environ. Adv. 4 2021 100042 Shruti V.C. Pérez-Guevara F. Elizalde-Martínez I. Kutralam-Muniasamy G. Reusable masks for COVID-19: a missing piece of the microplastic problem during the global health crisis Mar. Pollut. Bull. 161 2020 111777 Silva A.L.P. Prata J.C. Mouneyrac C. Barcelò D. Duarte A.C. Rocha-Santos T. Risks of Covid-19 face masks to wildlife: present and future research needs Sci. Total Environ. 148505 2021 Thiel M. de Veer D. Espinoza-Fuenzalida N.L. Espinoza C. Gallardo C. Hinojosa I.A. Kiessling T. Rojas J. Sanchez A. Sotomayor F. Vasquez N. COVID lessons from the global south–face masks invading tourist beaches and recommendations for the outdoor seasons Sci. Total Environ. 786 2021 147486 Wang Z. An C. Chen X. Lee K. Zhang B. Feng Q. Disposable masks release microplastics to the aqueous environment with exacerbation by natural weathering J. Hazard. Mater. 417 2021 126036

35090927

PMC9754081

NO-CC CODE

2022-12-16 23:26:26

no

Sci Total Environ. 2022 May 15; 821:153424

utf-8

Sci Total Environ

10.1016/j.scitotenv.2022.153424

oa_other

==== Front Lancet Lancet Lancet (London, England) 0140-6736 1474-547X Elsevier Ltd. S0140-6736(21)02250-9 10.1016/S0140-6736(21)02250-9 Correspondence Restoring biodiversity and slowing climate change are crucial to protect health Butler Colin D a Jaakkola Jouni J K bc Boylan Sinead d McFarlane Rosemary A e Potter John D fgh a National Centre for Epidemiology and Population Health, Australian National University, Canberra, ACT 0200, Australia b Center for Environmental and Respiratory Health Research, University of Oulu, Oulu, Finland c Finnish Meteorological Institute, Helsinki, Finland d Charles Perkins Centre and School of Life and Environmental Sciences, Faculty of Science, University of Sydney, Sydney, NSW, Australia e Faculty of Health, University of Canberra, Canberra, ACT, Australia f Research Centre for Hauora and Health, Massey University Wellington Campus, Wellington, New Zealand g Division of Public Health Sciences, Fred Hutchinson Cancer Research Center, Seattle, WA, USA h Department of Epidemiology, School of Public Health, University of Washington, Seattle, WA, USA 3 11 2021 13 11 2021 3 11 2021 398 10313 18021802 © 2021 Elsevier Ltd. All rights reserved. 2021 Elsevier Ltd Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. ==== Body pmcLukoye Atwoli and colleagues1 deliver a compelling call to address interacting global crises and to improve equity. Crucially, they link biodiversity loss with health and clearly warn that the Earth system is now too close to multiple tipping points, beyond which lie “catastrophic, runaway environmental change”.2 However, we think there is risk that the part of Atwoli and colleagues’ Comment concerned with future food security could give rise to pessimism. Irrespective of whether crop yield potentials are actually declining, food insecurity is deepening as a result of the effects of the COVID-19 pandemic and as a result of persisting unequal distribution of food and other forms of wealth and human rights. Scientists increasingly warn of synchronous or consecutive crop failures in multiple regions that produce large quantities of wheat and other food staples.3 Without transformational reform in global thinking, such a scenario is likely to exacerbate inequitable food aid, mirroring the world's self-defeating and unfair COVAX roll-out of COVID-19 vaccines. Alleviative strategies not mentioned by Atwoli and colleagues include markedly reducing the current diversion of crops for use as animal feed and fuel, such as soy (more than 90%), maize, sugar cane, and palm oil.4 Wild fish need not be fed to farmed, fish and food waste can be reduced. To benefit environmental and human health, a reduction in the average consumption of animal products by humans is crucial, both globally and especially in middle-income and high-income countries.5 Insect farming is increasing and might offer a way to expand the supply of nutrients for human and animal consumption at a lower environmental cost, including a reduced need for arable land and water. The nutritional status of the global poor can be bolstered by improving access to water and sanitation. Finally, health and other benefits will accrue from increased family planning support and improved education, including in regions that are currently food insecure. We declare no competing interests. ==== Refs References 1 Atwoli L Baqui AH Benfield T Call for emergency action to limit global temperature increases, restore biodiversity, and protect health Lancet 398 2021 939 941 34496267 2 Butler CD Climate change, health and existential risks to civilization: a comprehensive review (1989–2013) Int J Environ Res Public Health 15 2018 2266 3 Cottrell RS Nash KL Halpern BS Food production shocks across land and sea Nat Sustain 2 2019 130 137 4 Potter JD Red and processed meat, and human and planetary health BMJ 357 2017 j2190 5 Semba RD de Pee S Kim B McKenzie S Nachman K Bloem MW Adoption of the ‘planetary health diet’ has different impacts on countries' greenhouse gas emissions Nat Food 1 2020 481 484

34741817

PMC9754084

NO-CC CODE

2022-12-16 23:26:26

no

Lancet. 2021 Nov 13; 398(10313):1802

utf-8

Lancet

10.1016/S0140-6736(21)02250-9

oa_other

==== Front Vaccine Vaccine Vaccine 0264-410X 1873-2518 The Authors. Published by Elsevier Ltd. S0264-410X(22)00226-2 10.1016/j.vaccine.2022.02.073 Article The immunization Agenda 2030: A vision of global impact, reaching all, grounded in the realities of a changing world O'Brien Katherine L. a Lemango Ephrem b Nandy Robin b Lindstrand Ann a⁎ a Department of Immunizations, Vaccines and Biologicals, World Health Organization, Geneva, Switzerland b Health Section, Program Division, United Nations Children's Fund, NY, USA ⁎ Corresponding author. 15 12 2022 15 12 2022 © 2022 The Authors 2022 Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. Keywords Immunisation Strategy Vision Equity COVID-19 Vaccination ==== Body pmcThe launch of the Immunization Agenda 2030 (IA2030) came at an extraordinary and turbulent time in our collective experience. The COVID-19 pandemic erupted into all our lives, stopping dead in their tracks economies, education systems, human connections and many lives – lives of persons we loved. It has brought vaccines and immunizations back into the spotlight. In a single year we witnessed the emergence of a novel pathogen, the collective sprint to design, develop, evaluate, and manufacture numerous successful vaccines and deliver hundreds of millions of doses worldwide. This is a testament not only to the power, ingenuity and shared responsibility of collaborative science, but of our shared humanity. The acute phase of the pandemic will be controlled and a return to some sort of new normal will be achieved. COVID-19 vaccines are expected to make a major contribution toward that goal, and this remarkable achievement of vaccine science will need to be accompanied by continued political commitment, global economic partnership, and public engagement. At time of writing, the longer-term trajectory of SARS CoV-2 and the vaccine strategies to constrain its full impact are still unfolding, yet it is clear that the global vaccine ecosystem, which touches every community of every country around the world is being put to the test. Vaccine programs in the late 20th century and early 21st have contributed to the receding of infections previously ravaging human populations. Yet stubborn stagnation of vaccine coverage during the last decade, and severe backsliding during the pandemic years, has meant their full value is yet to be realized. Polio is stubbornly clinging to the last corners of the world, resurgent measles epidemics in countries of all means have taken hundreds of thousands of lives, and life-threatening diseases like diphtheria, meningitis, pneumonia, and diarrhea remain a reality for too many. This is a stark reminder that we cannot be lulled into a false sense of complacency about the need for vaccination and that achieving high coverage, with equity is anything but routine. We have to leverage this momentum to not only use vaccines as a critical tool to address the pandemic but to ensure equitable coverage of all available vaccines to prevent outbreaks and address health threats for individuals, in every community. The impact of vaccines to reduce morbidity and mortality in a highly cost-effective way is well quantified and well known. Their true impact can only be realized if they are accessible to everyone, particularly those who need them most. The IA2030 therefore, is a vision for our present and future world in which vaccines are available and accessible and provide protection from disease, fostering health and making possible life's opportunities to all, no matter the vagaries of birth, geography or economy. It is a vision co-created through iterative input from and representation of wide segments of the global community, stakeholders and end beneficiaries. A world where everyone, everywhere, at every age, fully benefits from vaccines for good health and wellbeing. This vision articulates a goal of reduced mortality and morbidity from vaccine-preventable diseases for everyone – infants, children, adolescents, adults and the elderly. A world in which no one is left behind, and access and use of new and existing vaccines is equitable. Where immunization systems build up, build out, and nest firmly within comprehensive primary health care and contribute to universal health coverage and sustainable development. These goals are grounded in real world strategic priorities and our experience in operating immunization programmes over many decades; it is these that are laid out and explicated in this Supplement. All seven IA2030 strategic priorities are essential building blocks to deliver the COVID-19 vaccines. The first of the IA2030 strategic priorities is that immunization programmes strengthen Primary Health Care and support Universal Health Coverage, which have been a focus of global public health ever since the Declaration of Alma Ata in 1978. The moment for unwavering political commitment towards this goal has never been stronger. This is achieved by reinforcing and sustaining strong leadership, management and coordination of high quality immunization programmes at all levels, and crucially by planning for and ensuring the availability of an adequate, effective, sustainable health workforce. It further requires comprehensive surveillance for vaccine-preventable disease including reliable, quality-assured laboratory networks with the right equipment, supplies, and expertise in place around the world. It requires robust supply chains for vaccines and related commodities and effective vaccine management, within the primary health care supply system. All aspects of immunization programmes must be underpinned by health information systems that provide high-quality data tailored for action for end-users, at all levels. A fundamental role of integration of primary health care, surveillance, data systems, planning and monitoring is the establishment and maintenance of vaccine safety surveillance, involving all stakeholders, and critical for trust in the health system, immunization programme and vaccine acceptance. The second strategic priority relates to political commitment & community demand for vaccination. Sustaining strong political and financial commitment for immunization is conceived broadly, with engagement at national and subnational levels of governance and close engagement with community groups, partners, stakeholders, vaccine advocates and vaccine recipients and their families. It is reflected in policy planning and in fiscal instruments. Decision making is to be better informed by evidence, with technical support from local expertise, such as national immunization technical advisory groups. Decision makers, stakeholders and all elements of the vaccine delivery chain need to be accountable for their part in the immunization programme, and data transparently reported to communities and civil society for joint monitoring. Additionally, through this strong community involvement and engagement, it will be important to understand social drivers of vaccine uptake and confidence, including social processes, gender-related barriers, practical factors and countering mis- or disinformation. Engagement, social listening, accountability and data transparency together support improving public trust and confidence as well as improving the quality of services, including convenience for the beneficiaries. Transparent and regular two-way communication of evidence will be important to counter disinformation, while still allowing open discourse, expression of public concern, and opportunities to address them by providing reassurances based on science and evidence. All these principles are being put thoroughly to the test as the global community and national governments work to deliver a constrained supply of COVID-19 vaccines to precipitously high but precarious demand, by listening to and addressing genuine community concerns through transparent vaccine safety systems and real-time data. By identifying and tackling barriers to procurement and fair allocation. As a global community we must succeed in using vaccines safely, effectively and equitably to deal with the current global crisis. And once we do, then vaccination programs should be leveraged to maintain political commitment of sufficient and sustainable resources, and maintaining high community confidence, demand and engagement. The third strategic priority is to achieve equitable vaccine coverage. This requires ensuring immunization services reach under-vaccinated or un-vaccinated children, the so-called ‘zero-dose’ children, and their communities. Data suggest that zero-dose children are largely clustered in remote rural, urban slum or conflict affected communities. These communities, which often face multiple deprivations will require context specific, tailored delivery strategies, requiring political engagement, specific investments, and a recognition that counting the previously uncounted may initially adversely affect coverage estimates. COVID-19 disruptions have widened immunization inequities. These disruptions need to be urgently addressed even while the programmes deliver COVID-19 vaccines at an unprecedented scale and speed. Much of our evidence and knowledge in immunization equity can be leveraged to ensure equitable distribution of COVID-19 vaccines – including for underserved populations like migrants, refugees and internally displaced populations in fragile conflict affected communities. The IA2030 prioritizes setting goals that are aspirational, but grounded in honest realities, rather than targets that are inspiring but simply cannot be met. High coverage should be achieved across all districts and populations, on the basis of values of justice, but also as a public good. Effort is therefore needed to identify and address barriers to coverage throughout the life-course especially among the most disadvantaged individuals and communities. Focusing on age, gender, location, social or cultural factors, allows development of evidence-based approaches to overcome barriers and achieving high, equitable coverage. The experience of disease eradication and elimination initiatives in reaching the most marginalized populations, and integrating immunization with disease control perspectives and with universal primary healthcare provisions will be important for reaching the under-vaccinated consistently and sustainably. Locally derived knowledge and ingenuity should be encouraged in order to develop context-specific, people-centered approaches to redress inequitable coverage. Central to any success of IA2030 are gender considerations which impact on programme leadership, design, community engagement, and most importantly household decision making on accessing vaccination services for the benefit of children, adolescents, pregnant women, and their families. The fourth strategic priority is to consider the role of vaccination throughout the life-course and integrated in health service delivery for people. This includes considering not only adherence to scheduled timepoints for infant vaccination but making easily available catch-up vaccinations and booster doses and identifying and addressing missed opportunities to vaccinate under-immunized individuals. It implies the establishment of integrated delivery points of contact between immunization and other public health interventions for different target age groups. For example, by providing adult vaccination through adult health services or following acute care interactions. These activities further require integrated data systems like vaccine registers that can be self-accessed or in some settings provider-accessed. It requires evidence on disease burden among older age groups and on the potential of vaccines to decrease that burden, directly and indirectly. It requires raising awareness of the benefits of vaccination beyond early childhood, through adolescence and in priority adult groups such as pregnant women, health workers and older adults, and requires implementation evidence on delivery strategies that work. It requires links beyond the communicable disease directorates of ministries with the directorates of non-communicable disease for example. The Covid 19 vaccines delivery to health care workers provides a long needed focus on other vaccines such as hepatitis B, oral cholera, influenza and eventually preventive Ebola vaccination. COVID-19 vaccination for older populations and medical risk groups is an opportunity to leap forward in building immunization programs for adults. Cross-sectoral collaboration with public and private health services, and indeed beyond the health care sector to ensure integration of immunization into programmes such as for education, nutrition, water and sanitation, care of older people and women’s empowerment, emphasizing the reciprocal benefit to general health achieved through vaccination. Links to prevention of disease through a ‘one Health’ approach motivates a deeper engagement with the veterinary community and focus on zoonotic diseases. Achieving these broad areas of integration may require programmatic restructure or even legislative action. It will require monitoring vaccination coverage at different ages. The fifth strategic priority addresses how vaccines and immunizations deal with outbreaks and public health emergencies. Immunization systems must be prepared to detect and rapidly respond to outbreaks of vaccine preventable disease. The response is at least two-fold: rapidly deploying available vaccines to shut down the outbreak, while using the outbreak as an opportunity to address critical gaps in programme performance which had led to the outbreak in the first place. This after-action response is critical for ensuring global health security and resilient routine immunization services are in place in the wake of an outbreak in order to break the chain of repeated outbreaks and responses. Outbreaks of a vaccine preventable disease will trigger opportunities to address gaps in immunization programmes for all vaccine preventable diseases. The importance of delivering vaccines in a timely manner in acute and protracted humanitarian emergencies, which are projected to grow in number and size over the decade is paramount to the right of all people to be protected from vaccine preventable diseases. The outbreak response to new pathogens with new vaccines, as has been the focus of the immunization community during 2020, will continue to be a priority. During the Covid 19 pandemic all immunization capacities have been stretched to the limit, highlighting that while we fight the pandemic, immunization programmes, which are essential health services, are eroded. In the most contemporary sense, we need to assess the risk of measles and other outbreak prone vaccine preventable diseases and close immunity gaps. This will require cross-sectoral coordination, investment in local capacity to sustained integrated surveillance and provide comprehensive response that can report to universal standards but that are locally and contextually applicable. Two-way communication and community engagement will be critical to promote participation in decision making and identify gaps in service availability. This strategic priority was conceptualized and authored before the full impact of the COVID-19 pandemic and response was felt. Much has been learned in the past year in terms of successes and costs of the pandemic response, and with respect to resilience and weak points in routine immunization systems. The lessons learned in difficult times are crucial for self-awareness and for building strong systems that can not only withstand future shocks, but can absorb them and mitigate their profound societal impact. The sixth strategic priority is ensuring sustainable vaccine markets and supply chains. This requires ensuring sufficient financial resources for immunization programmes in all countries and building and maintaining healthy global markets. The COVID-19 experience has shown that market forces alone are insufficient to address the needs of the world and that without coordination, convening, and structures to align competing interests, equitable market access is not achieved. Increasing independence from aid mechanisms for many countries by increasing domestic immunization expenditure and resource allocation to achieve and sustain high coverage for all vaccines is a desired but challenging goal. It will require supply innovation balanced with affordability, so that vaccine access and supply is timely. Forecasting and planning for procurement so that nations' needs are met and that manufacturers are aware and plan for the expected demand, including through diversification of quality-assured suppliers and strong local regulatory oversight. Good governance, stewardship and accountability of financing will be critical. Rapid access during public health emergencies requires special mechanisms, as have been established in the pandemic of 2020, including the COVAX Facility, other regional procurement mechanisms, manufacturer partnerships, public-partner investments, and agreements on transparency of information have all been components of the current response. The seventh and final strategic priority is research innovation in vaccine technologies, in vaccine delivery and in supply chain and logistics. The work in this strategic priority will leverage the massive set of lessons from COVID-19 vaccine development, scaleup, and deployment especially on partnership, open sharing of new evidence, adaptive research methods, research collaboration and technology transfer. Developing improved vaccines for endemic diseases like influenza, TB, and malaria, as well as new vaccines for diseases as yet not vaccine preventable, like HIV will continue into this decade. Vaccines for previously unknown emerging pathogens like SARS-COV-2 and preparedness for a future ‘Disease X’ are central parts to this strategy. It also requires improving on existing vaccines, especially for populations in which achieving immunity is biologically challenging like newborns or the elderly. It requires sustained investment in promising ideas that have scalable capacity. Innovations in delivery should be mindful of prioritizing community needs, particularly for underserved populations. Real-world, evidence-based innovations in operational research can shorten the path to maximal and equitable vaccine impact. The IA2030 vision and priorities are founded on four core principles. (1) That strategy should be people centered, shaped by and responsive to lived realities, including the diversity of experience. It should prioritize achievable real-world impact over idealized goals. (2) It should be country owned, with progress driven and built from the ground up. Targets and milestones should be ambitious, yet with a feasible pathway to being a reality. Progress to these milestones should be transparent and linked to accountability. (3) Systems should be based on strong and diverse partnerships, coordinating multi-sectoral efforts to maximize impact. (4) Decision making should be guided by data and based on high-quality, “fit-for-purpose” evidence. Ongoing monitoring allows for coordinated course corrections in real time. Data must be available and used to drive programmatic changes, while reporting on progress toward targets must be honest and transparent, with health systems held accountable for such progress. The IA2030 is a vision and strategy for all. It is owned by all countries, all partners, and all communities. It was endorsed by all WHO Member States in November 2020, formally expressing the commitment of all country governments to the vision and the strategic components that will underpin its success. Its development has been collaborative and broad-based and reflects the views of practitioners and users across the globe. This participatory, co-created and articulated vision is both optimistic and realistic. IA2030 comes with implementational strategies and regions and countries are actively encouraged to tailor it to their priorities and changing needs. Its impact goals are ambitious and only reachable with an unwavering focus on reaching those most underserved, and unswerving commitment and partnership beyond the health sector, collaboratively to build stronger and more efficient health systems. Systems that reach all, and are responsive to issues of gender, age, and access across the life course. The IA2030 encourages innovation, responsiveness to local need through the transparent use of quality data to drive any assessment and any action to best tailor immunization programs. The IA2030 was being at a pivotal time when COVID-19 vaccines were driving the priorities, expectations and hopes of our global society. Through this pandemic we have come to recognize the weaknesses in global health security and the vulnerabilities of health systems. The IA2030 aims to show us the way to resilience, and the ability of integrated universal health systems to provide comprehensive immunization programs. Implementation of the IA2030 will better prepare us for the next pandemic. But more than this, IA2030 reflects a fundamental faith in our ability to work together, mutually, respectfully and collaboratively, building upon one another's expertise and perspectives. Together we really can build a world where all benefit from the advancements and miracle of modern vaccinology. Disclaimer: The authors alone are responsible for the views expressed in this article and they do not necessarily represent the views, decisions or policies of the institutions with which they are affiliated. Declaration of Competing Interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

0

PMC9754085

NO-CC CODE

2022-12-16 23:26:26

no

Vaccine. 2022 Dec 15; doi: 10.1016/j.vaccine.2022.02.073

utf-8

Vaccine

10.1016/j.vaccine.2022.02.073

oa_other

==== Front Lancet Lancet Lancet (London, England) 0140-6736 1474-547X Elsevier Ltd. S0140-6736(21)00992-2 10.1016/S0140-6736(21)00992-2 World Report Controversy surrounds Merkel's new lockdown powers Hyde Rob 29 4 2021 1-7 May 2021 29 4 2021 397 10285 16101610 © 2021 Elsevier Ltd. All rights reserved. 2021 Elsevier Ltd Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. ==== Body pmcFollowing a parliamentary victory, the German chancellor can force municipalities into lockdowns to curb COVID-19. Rob Hyde reports. Germany's 1949 constitution, created in the aftermath of World War 2 and designed to limit centralised government control, forged a republic by entrusting each region, or federal state, with far-reaching powers. 71 years later, Germany's states are semiautonomous, each manages key policy areas from school syllabuses to the provision of health care. The interplay between national and regional politics lies at the heart of Angela Merkel's newly won lockdown powers. According to the German constitution, the national government is responsible for taking ”measures against diseases that are dangerous to the public, or are transmissible”. However, up until last week, even under the terms of the government's Infection Protection Act (Infektionsschutzgesetz), the 16 German states still held powers to enact and enforce their own COVID-19 measures. The result was a patchwork of regional approaches to curfews, retail, schools, and sports. With skyrocketing infection rates, Merkel had urged the 16 regional leaders to back her push for tougher lockdowns across Germany. When this proved politically unviable, however, she changed tack. Turning to the German parliament, the Bundestag, she sought the legal right to force municipalities with over 100 cases per 100 000 people to activate a so-called emergency brake and enter lockdown. The move proved highly controversial. As Merkel's vote was held inside the Reichstag on April 21, 2021, around 10 000 protesters clashed with 2200 police officers outside the government buildings. 29 officers were injured, and 230 arrests were made. Inside the Reichstag, Merkel won: 342 German MPs voted for, 250 against, and 64 abstained. Merkel's Christian Democratic Union officially backed their leader's deal, as did the Bavarian sister party, the Christian Social Union, and the Social Democratic Party. In a written statement to The Lancet, Bärbel Bas, deputy chairperson of the Social Democratic Party parliamentary group for health, education, and research and petitions, said that by backing Merkel's deal, the Bundestag was “fulfilling its duty to protect citizens”. “We need an emergency brake…In the case of particularly high infection figures, clear and uniform federal regulations are needed that are easy for everyone to understand.” Although the Green Party (Die Grünen) abstained, Merkel's plan was fiercely opposed by both the far-left Die Linke party and the far-right Alternative für Deutschland party. The plan was also rejected by the liberal Free Democratic Party. In a statement to The Lancet, Andrew Ullmann, chairman of the health committee for the Free Democrats' parliamentary group, wrote Merkel's emergency brake risked unpicking Germany's entire federal system. “In Germany, the principle of federalism is deeply rooted in historical responsibility. We must not shake this. Finding unity in diversity must not lead us to abolish our federal constitutional state.” Merkel's emergency break has received a mixed blessing from Klaus Reinhardt, president of the German Medical Association. Despite saying that an emergency brake was “the right thing to prevent ICUs [intensive care units] from being overloaded”, Reinhardt said it was wrong to use case rates as the only criterion for determining lockdowns. He argued that figures on vaccinations delivered per day and numbers of COVID-19 patients in ICUs should also be used. Thomas Gerlinger, professor in the School of Public Health at Bielefeld University (Bielefeld, Germany), told The Lancet that Merkel's emergency brake could lead to “more clarity” in Germany because the situation up until now has been “totally chaotic”. “In Saxony-Anhalt, I could shop in a department store. But at the same time, in Berlin, the number of COVID cases per 100 000 was lower…Despite this, people had to present a negative [SARS-CoV-2] test result to enter Berlin's shops. Such conflicting examples breed public mistrust.” Others, however, feel Merkel's emergency brake is unlikely to help. Jens Holst, professor at the department of health sciences at Fulda University of Applied Sciences (Fulda, Germany), says that, before the recent changes, the regions had been well served by a diversified approach. “I don't think a blanket policy is the way forward. Some states need stricter and milder responses accordingly.” “It wasn't really a problem that regional laws were different before. Residents objecting to local [COVID-19] regulations could simply hop state and have a rather different experience.” This online publication has been corrected. The corrected version first appeared at thelancet.com on May 20, 2021

33933196

PMC9754086

NO-CC CODE

2022-12-16 23:26:26

no

Lancet. 2021 Apr 29 1-7 May; 397(10285):1610

utf-8

Lancet

10.1016/S0140-6736(21)00992-2

oa_other

==== Front Lancet Lancet Lancet (London, England) 0140-6736 1474-547X Elsevier Ltd. S0140-6736(21)00885-0 10.1016/S0140-6736(21)00885-0 Correspondence The Swedish COVID-19 strategy revisited Claeson Mariam a Hanson Stefan b a Results for Development, Washington, DC 20036, USA b Spånga Transkulturella Läkarmottagning, Spånga, Sweden 19 4 2021 1-7 May 2021 19 4 2021 397 10285 16191619 © 2021 Elsevier Ltd. All rights reserved. 2021 Elsevier Ltd Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. ==== Body pmcIn December, 2020, we wrote about the Swedish response to the COVID-19 pandemic.1 Our hope was that our Comment, together with hundreds of other fact-based articles, would gain the attention of the Swedish Public Health Agency (Folkhälsomyndigheten [FHM]), that they would revisit and change the national strategy that they had designed so that it would be more aligned with global best practice, and that the political decision makers would act on it. They did not. Since then, the FHM has recorded more than 5600 deaths from COVID-19 in Sweden, and cases and deaths continue to rise as we face the third wave without any widespread sense of gravity or urgency. The debate among critics of the Swedish national approach to the pandemic has been consistent since March, 2020: be strategic, test and trace more, follow the growing evidence base and recommend the use of face masks, and enforce regulations about physical distancing and ventilation, especially in schools if they are open. Some critics have advocated for more government-led legal interventions such as reinforcing quarantine or lockdown. It has been a call for timely implementation of basic principles of pandemic prevention and control to contain the spread and flatten the curves of hospitalisations, deaths, and chronic illness. Instead of following evolving evidence, the FHM has doubled down and defended its approach without reconsidering the assumptions on which the failed national approach is based. It has downplayed the roles of asymptomatic spread, aerosol transmission, children as potential source of infection, and the use of face masks. It has maintained an approach that mainly builds on recommendations to take voluntary actions, guided (in our view) more by public opinion than by sound public health policy. The media has played a crucial role in this pandemic response, mostly lacking in investigative journalism and failing to question or hold the public health agency accountable, with some exceptions.2 Dagens Nyheter, a major newspaper, recently exposed3 Sweden's large inequities in COVID-19 deaths across income, education, and origin of birth—data that should have informed the national strategy from its inception. As of April 16, 2021, more than 13 700 people have died from COVID-19 in Sweden. The country has one of the highest infection rates in western Europe according to Our World in Data COVID-19 statistics, with 606 new infections per million per day, while its neighbours Denmark, Finland, and Norway reported 115, 62, and 112 new infections per million per day, respectively (April 15, 2021). New and more infective and deadly variants have taken over, and by April 15, 2021, the UK SARS-Cov-2 variant was supected to have caused 75–100% of all new cases in all regions. This indicates more rapid spread, more deaths, and that more young people will be affected, with intensive care units already at full capacity in some regions.4 While other countries are closing down in response to this new surge in cases, Sweden is opening up—high schools were opened on April 1, 2021. To continue on the same trajectory in the face of current trends, without timely action by agency and government leadership, raises concerns about governance and accountability, and ultimately about fundamental ethics and values. © 2021 Carl-Olof Zimmerman/Getty Images 2021 We declare no competing interests. ==== Refs References 1 Claeson M Hanson S COVID-19 and the Swedish enigma Lancet 397 2020 259 261 33357494 2 Ridderstedt M Öhman D Coronapandemin: den andra vågen, del 1 www.sverigesradio.se/avsnitt/1706454 April 13, 2021 3 Dagens Nyheter Unik kartläggning: här är de som dött I covid i Sverige www.dn.se/sverige/unik-kartlaggning-har-ar-de-som-dott-i-covid-i-sverige/ March 22, 2021 4 Region Uppsala Ökad spriding av covid-19 ger extrem belastning på vården. Uppsala https://via.tt.se/pressmeddelande/okad-spridning-av-covid-19-ger-extrem-belastning-pa-varden?publisherId=3235664&releaseId=3296545 April 6, 2021

33887212

PMC9754087

NO-CC CODE

2022-12-16 23:26:26

no

Lancet. 2021 Apr 19 1-7 May; 397(10285):1619

utf-8

Lancet

10.1016/S0140-6736(21)00885-0

oa_other

==== Front Lancet Lancet Lancet (London, England) 0140-6736 1474-547X Elsevier Ltd. S0140-6736(21)00940-5 10.1016/S0140-6736(21)00940-5 Perspectives Comics as anti-racist education and advocacy Obuobi Shirlene a Vela Monica B a Callender Brian ab a Department of Medicine, University of Chicago, Chicago, IL 60637, USA b Stevanovich Institute on the Formation of Knowledge, University of Chicago, Chicago, IL 60637, USA 29 4 2021 1-7 May 2021 29 4 2021 397 10285 16151617 © 2021 Elsevier Ltd. All rights reserved. 2021 Elsevier Ltd Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. ==== Body pmcAcademic medicine is increasingly recognising the importance of teaching about structural racism in medicine to help ameliorate racial health-care disparities. Yet such teaching can be challenging and, in some settings, considered controversial. Leveraging the power of narrative, comics can contribute to education about structural racism. Structural racism involves the normalisation and proliferation of inequitable and interconnected societal systems, policies, institutions, ideologies, and practices that disadvantage, discriminate against, and reinforce inequities faced by racialised minorities. In the broader societal context of the USA, examples of structural racism include discriminatory lending practices that continue to bar Black, Indigenous, and people of colour from home ownership and access to quality education and initiatives that place the burden of harmful environmental exposures on minoritised neighbourhoods or limit access to public transportation, public spaces, voting rights, and healthy food options. In the history of medicine, structural racism is apparent in a legacy of experimentation on Black bodies, colonial and racialised medicine, “scientific” racism, and the segregation of hospitals. In contemporary health care, it includes the persistence of racialised medicine and science, unequal access to health care, clinical training programme ranking systems that disadvantage minority students, persistently disparate outcomes in Black maternal and infant mortality and any number of health and health-care disparities, and the continuing under-representation of communities of colour in academic medicine, health-care leadership, research, and on the boards of health-care organisations. The stark racial disparities in COVID-19 cases, morbidity, mortality, and vaccination rates have amplified awareness and discussion of structural racism. Despite increased attention on the topic, the past year has been fraught with incidents that highlight how entrenched structural racism is in medicine and how challenging it can be to address. Despite much scholarly literature that espouses the importance of addressing structural racism to reduce health inequities, medical institutions often struggle to fit impactful teaching about it into curricula or even dismiss physician educators who teach about structural racism or advocate against biased policies. Such discomfort with engaging in discourse about structural racism makes teaching the topic all the more necessary. In recent years, there have been calls to action for anti-racism courses in medical education to draw attention to and help dismantle these structures. This coursework should include a structural competency curriculum with in-depth study of structural determinants that impact the health and wellbeing of populations, an opportunity for transformational learning that provides for the growth of critical consciousness, and the role modelling of engagement in advocacy at every level from individual patient advocacy to system-level changes. Teaching about structural racism requires thoughtful, effective forms of communication. Peek and colleagues, for example, outline deliberate measures educators should take to create a psychologically safe space for conversations about racism, noting that it can help to “start with stories, not numbers”. This approach, which encourages learners to engage with empathy, can help diffuse tension and refocus them on a likely shared goal of improving outcomes for all. Approaches include using videos showcasing instances of interpersonal racism, curricula involving small group discussions that allow for intimate conversations about injustice, and lectures by academic experts in the subject, such as social scientists and medical historians. Unsurprisingly, some successful modes of education incorporate reading and creating narratives because they help to imbue lessons with humanity, making them relatable and more accessible. As a readily recognisable form of narrative, comics are a powerful tool in anti-racist education and advocacy. Comics combine textual and visual elements to express ideas and create narratives. In our increasingly fast-paced world, comics can succinctly communicate information and evoke emotional responses, often by simultaneously depicting multiple narratives and balancing levity with seriousness. In this way, comics can draw attention to medical and societal inequities that might otherwise be normalised by displaying contrasting narratives in parallel panels (Figure 2, Figure 2 ). The directness with which comics engage a reader forces one to contend with the topics, the characters, and the story. However, given that the interpretation of the visuals in comics is left up to the viewer, they also create opportunities for personal interpretation and interpersonal discussions. These discussions remind readers of the existence of perspectives different from their own, and the fact that although readers can all see the same thing, they can interpret and experience it differently.Figure 1 Comparison between health-care providers' attitudes to patients with cystic fibrosis and sickle cell disease CF and SCD are two autosomal recessive conditions with disease profiles notable for serious complications in childhood and early adulthood causing reduced life expectancies. CF is more common in populations with northern European ancestry and SCD is more common in populations with west African ancestry. Despite the disease similarities and the fact that SCD is about three times more prevalent in the USA, US National Institutes of Health funding for CF-based research is higher than that allocated to SCD-based research. Bias against patients with SCD seeking treatment with opioids has been well explored in the literature, and attitudes surrounding the treatment of this population by the health-care community can be affected by bias. This comic displays the parallel narratives of two similar diseases to expose the insidious but powerful impact of systemic racism in our approaches to patient care. CF=cystic fibrosis. SCD=sickle cell disease. © 2021 Shirlene Obuobi 2021 Figure 2 Discussions about racial disparities dodge structural racism This comic addresses the readiness with which medical education may circumvent conversations about the role of structural racism in health-care disparities. By not addressing structural racism as a root cause for these disparities, medical trainees may infer supposed biological differences. One study found that a considerable proportion of medical students surveyed in the USA believed that Black patients were less sensitive to pain or had thicker skin than White patients; these false beliefs may lead to biases when caring for Black patients, leading to racial disparities in patient care. © 2021 Shirlene Obuobi 2021 Many comic artists have already taken advantage of these qualities to address racism, including John Jennings, co-editor of The Blacker the Ink: Constructions of the Black Identity in Comics and Sequential Art, an award-winning collection of comics on Black identity and structural racism, and Whit Taylor, whose comics address such topics as increased mortality and morbidity faced by Black mothers (Black Mothers Face Far Worse Health Outcomes. How Do We Fix It?) and distrust in the medical system (African-Americans are More Likely to Distrust the Medical System. Blame the Tuskegee Experiment). Within medicine, comics have long been recognised for their ability to depict the illness experience, the trials and tribulations of medical education, and the challenges of health-care delivery. More recently, however, some medical establishments and health professionals have used comics to address racism and health disparities. In A Sense of Belonging, a comic published in The New England Journal of Medicine, physician Anita Blanchard discusses how racial disparities are driven by generational privilege, calling it the “oldest form of ‘affirmative action'” which “continues to challenge efforts to create a level playing field for physicians from minorities that are underrepresented in medicine”. This point is carried forward with the background of her own journey into medicine, entering medical school as a hopeful and eager student excited to care for her community, only to find that she was already behind her classmates. Although not a focus of their comics, physician artists such as James Fulmer and Michael Natter have used the medium to address topics such as inadequate interventions for racial disparities in health care and the importance of directly naming and addressing structural racism. The social media platform of physician artist Shirlene Obuobi, who is a co-author of this piece, regularly explores topics related to structural racism in medicine. Under the moniker ShirlyWhirl, M.D. (@shirlywhirlmd), her comics reflect on her personal career path, following her through medical school and beyond. Initially, many of those reflections faced inward and portrayed such challenges as mastering difficult medical concepts, unfair evaluations, and attending physicians who did not remember her name. In residency, however, her comics have become more outward looking and explore such topics as physician workflow, the difficulties of providing equitable patient care under the USA's current health insurance model, sexual harassment, physician wellness, and structural racism. ShirlyWhirl, M.D's comic narratives serve as a way to bear witness to the system that provides patient care and to advocate and spread awareness of dysfunctional but normalised structures. Although these comics do not exhaustively delve into the scholarly literature on these topics, they offer a visual glimpse that seeks to inspire curiosity and engagement rather than reticence and avoidance. Colour, style, and space are used in contrasting ways in these narratives to create mood. The general colour scheme of ShirlyWhirl, M.D. comics is bright and colourful. However, occasionally, the comics are longer, with the narrative stretched across multiple panels and presented in black and white (figure 3 ). Viewers implicitly understand this transition when approaching the comics. They may expect irreverence when reading the colourful, short comics, but stark emotion when reading the black-and-white narratives.Figure 3 Excerpt from Whistling, June, 2020 This comic, drawn shortly after George Floyd's murder in 2020, showcases an image of the artist holding her imagined future children, who are filled in greyscale, while the ghostly figures of Black children who have been killed by police in the USA look on. Panels like this one take advantage of temporal devices that can be used in comics; this panel simultaneously depicts the past, the present in the textual expression of ShirlyWhirl, M.D's fears, and the future. In addition, the figures are only viewed from behind, making the audience extrapolate on their expressions and add their own emotional contexts to the scene. Panels such as this one take advantage of perspective. Most people can empathise with fearing for one's children, and the demonstration that fear of police brutality, which is abetted and often left unpunished due to structural racism, is something the artist has considered when thinking of starting a family. © 2021 Shirlene Obuobi 2021 The main character of ShirlyWhirl, M.D. is an avatar of Shirlene, a young, Black female physician who frequently code switches between standard English and terms common in African American Vernacular English. This depiction contrasts with more traditional images of physicians, who are usually White, male, middle-aged, and wielding medical jargon like an intellectual baton. By repeatedly exposing readers to her portrayal of a physician, ShirlyWhirl, M.D.'s work expands perceptions of what a doctor should look and sound like. Another regular in ShirlyWhirl, M.D.'s comic series is a purple, featureless, humanoid figure who has come to represent health-care providers, administrations, and institutions (Figure 2, Figure 2). The identity of this figure is never clearly stated, allowing the viewer to extrapolate and project upon them an identity they may deem fits the role. By using this recurring figure, who is both alien and familiar, the comic draws focus to the system or practices being critiqued that are instrumental in perpetuating structural racism and systemic inequities. ShirlyWhirl, M.D.'s comics are intentionally shared primarily through social media platforms that are accessible to people inside and outside of medicine, and thus the comics are read by health-care workers and laypersons alike. When asked about the role of comics in expanding their knowledge of and comfort with topics related to structural racism, readers of ShirlyWhirl, M.D.'s comics have said that they found comics easier to read and process than denser formats such as news articles or academic studies or publications. They thought comics were less “hostile” than other, potentially less accessible media, and some said that specific comics had allowed them to initiate conversations about racism with otherwise reticent loved ones. By making her comics readily available so that they can be discussed on an open forum, ShirlyWhirl, M.D. creates space that can feel safe for these conversations. Taken together, comics that deal with racism in medicine and health care are a useful component of education about and advocacy against structural racism. These narratives are powerful expressions of the lived experience of the social and structural effects of racism and allow for more inclusive discussions to confront this reality and promote anti-racist actions. Leading with stories and incorporating narratives about racism, whether through comics or other narratives and expressive media, is an important pedagogical tool to engage learners and create a safe educational environment. However, these stories about racism have a broader relevance beyond the classroom. Collectively, they assert a social and cultural reality that demands more than acknowledgment and discussion; this is a collective experience that demands change to the structures and systems within medicine and society that perpetuate racism and racial health disparities. We hope that the stories now being told will be part of the meta-narrative that reflects an arc towards justice and equality within medicine and health care. ==== Refs Further reading Bailey ZD Krieger N Agénor M Graves J Linos N Bassett MT Structural racism and health inequities in the USA: evidence and interventions Lancet 389 2017 1453 1463 28402827 Blanchard A Koscal N Burke AE A sense of belonging N Engl J Med 383 2020 1409 1411 33027569 Green MJ Myers KR Graphic medicine: use of comics in medical education and patient care BMJ 340 2010 c863 20200064 Nuriddin A Mooney G White AIR Reckoning with histories of medical racism and violence in the USA Lancet 396 2020 949 951 33010829 Cooper Owens D Listening to Black women saves Black lives Lancet 397 2021 788 799 33640056 US Centers for Disease Control and Prevention Data and statistics on sickle cell disease https://www.cdc.gov/ncbddd/sicklecell/data.html US Centers for Disease Control and Prevention Cystic fibrosis https://www.cdc.gov/genomics/disease/cystic_fibrosis.htm Dao DK Goss AL Hoekzema AS Integrating theory, content, and method to foster critical consciousness in medical students: a comprehensive model for cultural competence training Acad Med 92 2017 335 344 27680318 Farooq F Mogayzel PJ Lanzkron S Haywood C Strouse JJ Comparison of US federal and foundation funding of research for sickle cell disease and cystic fibrosis and factors associated with research productivity JAMA Netw Open 3 2020 e201737 Hoffman KM Trawalter S Axt JR Oliver MN Racial bias in pain assessment and treatment recommendations, and false beliefs about biological differences between blacks and whites Proc Natl Acad Sci USA 113 2016 4296 4301 27044069 Lubeck D Agodoa I Bhakta N Estimated life expectancy and income of patients with sickle cell disease compared with those without sickle cell disease JAMA Network Open 2 2019 e1915374 Metzl JM Hansen H Structural competency: theorizing a new medical engagement with stigma and inequality Soc Sci Med 103 2014 126 133 24507917 Peek ME Vela MB Chin MH Practical lessons for teaching about race and racism: successfully leading free, frank, and fearless discussions Acad Med 95 2020 S139 S144 32889939 Ray KS Going beyond the data: using testimonies to humanize pedagogy on Black health J Med Humanit 2021 published online Feb 12. 10.1007/s10912-021-09681-7 Smith LA Oyeku SO Homer C Zuckerman B Sickle cell disease: a question of equity and quality Pediatrics 117 2006 1763 1770 16651336 Whitehead-Clarke T More on racial bias in pulse oximetry measurement N Engl J Med 384 2021 1278 White-Davis T Edgoose J Brown Speights JS Addressing racism in medical education: an interactive training module Fam Med 50 2018 364 368 29762795

33933198

PMC9754088

NO-CC CODE

2022-12-16 23:26:26

no

Lancet. 2021 Apr 29 1-7 May; 397(10285):1615-1617

utf-8

Lancet

10.1016/S0140-6736(21)00940-5

oa_other

==== Front Lancet Lancet Lancet (London, England) 0140-6736 1474-547X Elsevier Ltd. S0140-6736(21)00993-4 10.1016/S0140-6736(21)00993-4 World Report Experts criticise India's complacency over COVID-19 Bhuyan Anoo 29 4 2021 1-7 May 2021 29 4 2021 397 10285 16111612 © 2021 Elsevier Ltd. All rights reserved. 2021 Elsevier Ltd Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. ==== Body pmcMass gatherings have been permitted as cases soar and patients die, while experts criticise a lack of planning and flexibility in the COVID-19 response. Anoo Bhuyan reports from New Delhi. India is battling a second wave of COVID-19, which has rapidly surpassed its first wave in 2020 in terms of the number of new cases and deaths per day. Currently, India has the second highest number of COVID-19 cases in the world after the USA. “The country is working day and night for hospitals, ventilators, and medicines”, said India's Prime Minister in his monthly national broadcast on April 25, 2021. India has been recording more than 300 000 cases of COVID-19 per day since April 21, up from 100 000 per day on April 4. These numbers eclipse India's previous highest number of new cases reported in a single day, at 97 860 cases on Sept 16, 2020. Health infrastructure has collapsed in several cities, with several state governments imposing curfews and lockdowns on movement, such as in the national capital New Delhi and in Maharashtra. State governments are scrambling to build up new infrastructure, making announcements this month about suddenly commencing the construction of new health facilities or oxygen plants. However, this frenetic activity comes in the middle of an ongoing and exponential rise in cases, whereas it should have come before, say experts. In early 2021, an opinion that India had overcome the pandemic and acquired herd immunity gained ground among policy makers, sections of the media, and the public, said Srinath Reddy, president of the Public Health Foundation of India. “Even sections of the scientific community propagated this view”, he added. The belief that there would be no second wave, says Reddy, was also spurred on by the desire to reopen society and revive economic growth. Although India saw a lull in cases in January and February of this year, March was a period of hectic public gatherings, sanctioned and even encouraged by public officials. Five states held elections this month, and many politicians, including India's prime minister and leaders of several parties, conducted hundreds of massive political rallies around India. Just 10 days ago, in an address at an election rally in West Bengal, India's Prime Minister Narendra Modi said he had never seen such huge crowds at a rally. Last month, he tweeted: “On my way to the massive party rally.” The Bharatiya Janata Party, to which he belongs, has been regularly publishing the location and timing of his various public rallies for people to attend. India's election commission, responsible for organising all elections, had repeatedly published notices threatening to act against politicians for their massive rallies and roadshows during the pandemic, but these notices have not amounted to much as the commission has not taken action against any political party for the crowded rallies. Only last week did they finally take their strongest action yet of banning political roadshows. However, the commission has still permitted public meetings by politicians, with the caveat that they be kept to under 500 attendees. Despite the pandemic and the risk of a major rise in cases, central and state governments also permitted the Hindu festival of Kumbh Mela to go ahead. Millions of Hindus turned up to the festival for prayers and a dip in the river Ganges, which is considered auspicious. The festival began on April 1, and was only called off by local authorities 17 days later. Local authorities reported nearly 2000 cases of COVID-19 detected among people who had come to participate in the festival. Fully opening society with unrestrained crowding, mass gatherings, large scale travel, and lack of personal protective measures such as masks “permitted the virus to move freely”, said Reddy. “Large mass gatherings should have been avoided”, he said. This action could have not only protected participants in these mass gatherings, but also prevented others from getting the wrong signal that the danger had fully passed. The most palpable and visceral crisis in the country currently is a shortage of oxygen in hospitals. “We are delivering oxygen cylinders to people's houses when they call us in an emergency”, said Harteerath Singh, a volunteer with Hemkunt Foundation in Delhi. Singh and a group of volunteers have about 200 oxygen cylinders, which they constantly get refilled from various vendors and deliver to people's homes around Delhi. Singh said they are currently finding it difficult to get any new cylinders or refill the ones they have because vendors everywhere are busy or out of stock themselves. Indian social media is awash with thousands of requests from all over the country of people asking if there are any oxygen cylinders or hospital beds with oxygen or ventilators available. India's daily production capacity for oxygen is 7127 metric tonnes and consumption is 3842 metric tonnes, according to Indian Government data released in early April. However a few days later, Max Hospital, a private hospital, approached a Delhi court to inform them about an oxygen shortage at their facility. During the hearing, the government is reported to have told the court that India's oxygen consumption was over 8000 metric tonnes per day by April 21. © 2021 Sonu Mehta/Hindustan Times/Getty Images 2021 “I dread receiving calls from family or friends these days as mostly it is to seek help in finding a bed. In most cases I have failed”, Tweeted Indu Bhushan, a senior bureaucrat who was instrumental in setting up India's massive health insurance scheme in 2018. The central and state governments had scaled back some of the arrangements they had made for oxygen in hospitals after the first wave subsided. As COVID-19 numbers were waning, perhaps this was alright, said T Sundararaman, former dean of the School of Health System Studies at the Tata Institute of Social Sciences, but there was no “flexibility” in the system to ramp it up quickly as numbers rose again, he said. “The government did not plan for both peak and non-peak scenarios”, he said, which explains why there is not enough supply to meet the demand as India has hit yet another and a much higher peak. “The government needs to take responsibility for both production and logistics of distribution of this oxygen and cannot claim that there is enough supply or that supply is now being created, without also handling its distribution.” The Indian Government only announced on April 15 that it is looking to import 50 000 metric tonnes of medical oxygen. It has asked its foreign missions in different countries to find vendors who can supply medical oxygen to India. The EU will “do its utmost to mobilise assistance”, especially with regard to oxygen supply and medicines, according to Janez Lenarcic, the European Commissioner for Crisis Management. Similar messages of support have been made by officials in Germany and France. The Indian Government has also sent aircrafts to Singapore to airlift liquid oxygen containers and deliver them to India. The US Government issued a statement on April 25, 2021, saying that it will release raw materials that are needed for making COVID-19 vaccines to India, and that it is looking into how to send oxygen generation equipment “on an urgent basis”. However, the higher number of COVID-19 cases in India has also triggered foreign governments to be more cautious. The UK Government has banned visitors travelling from India from entering the UK, and the French Government has imposed a 10-day quarantine on any traveller from India entering France. The Indian Government has said that India's programme to give the COVID-19 vaccine is the “world's largest”. Indian companies are also manufacturing and exporting COVID-19 vaccines to the rest of the world. For example, Serum Institute of India is exporting the AstraZeneca COVID-19 vaccine, and several Indian companies are also manufacturing the Sputnik-V vaccine. However, despite manufacturing vaccines for other countries, India is facing a shortage of vaccines for its own programme. Some people who have received their first injection of the two vaccines in use in India (Covaxin and Covishield) have been unable to get their second dose as vaccination centres around the country are reporting an absence of replenishments. “It was all bad planning”, said Shahid Jameel, a virologist at Ashoka University in New Delhi. “India did not give sufficient orders to vaccine companies, to allow them to manufacture enough doses. Whereas other countries, which are taking vaccination seriously, had all given assured orders to vaccine manufacturers.” The Indian Government last declared on April 8 how much vaccine stock it had, when the health minister reported a stock of 24 million doses. As of April 26, the country has administered around 145 million doses, and has said that it intends to administer 500 million by July, 2021. However, the government has not given any data as to where and when the millions of vaccine shots needed for the targeted groups would be available. All the same, the government has announced that all adults older than 18 years will be able to get a COVID-19 vaccine, starting May 1, 2021. Despite this shortage, India is also continuing to export vaccines made commercially in India as donations and to WHO's COVAX Facility. “India's vaccine diplomacy and policy of exporting and donating vaccines was a good thing”, said Jameel. “But we underestimated our demand.”

33933197

PMC9754089

NO-CC CODE

2022-12-16 23:26:26

no

Lancet. 2021 Apr 29 1-7 May; 397(10285):1611-1612

utf-8

Lancet

10.1016/S0140-6736(21)00993-4

oa_other

==== Front Lancet Lancet Lancet (London, England) 0140-6736 1474-547X Elsevier Ltd. S0140-6736(21)00713-3 10.1016/S0140-6736(21)00713-3 Correspondence Global health and its discontents Rasanathan Kumanan a a Health Systems Global, 12302 Phnom Penh, Cambodia 22 4 2021 24-30 April 2021 22 4 2021 397 10284 15431544 © 2021 Elsevier Ltd. All rights reserved. 2021 Elsevier Ltd Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. ==== Body pmcA year ago, a group of us gathered to reconsider how to build healthy and equitable societies.1 During that meeting, we rehearsed critiques of the current practice of global health.2 The COVID-19 pandemic has laid bare the truth in these charges, highlighting deficiencies in the pursuit of equity and in the capacity for multisectoral action—yet the pandemic has also provided inspiring examples of effective national and global public health action. After the 1918 influenza pandemic, many countries built new institutions, laws, and practices that laid the foundation for modern public health. As a global health community, we should not miss the opportunity from this crisis to reflect upon and remedy our shortcomings to better support global health equity. First, despite genuine desire and goodwill, and the Paris and Accra declarations, global health remains insufficiently country-centred. COVID-19 has provoked impressive examples of global solidarity, but it has also shown that individual decisions at the national level matter more for health than regional and multilateral institutions and mechanisms, including global health treaties and strategies. Global health practice often fails to fully engage with the individual context, policy cycles, and political economy of national health and social systems. Too many global technical and policy documents provide insufficient direct guidance and detail for national decision makers or are pitched at the wrong level. The unpredictable, short-term, feast-or-famine nature of overseas development assistance for health has proved difficult to improve. Yet navigating the sometimes conflicting internal and external incentives that give rise to this incoherence remains essential to improving the utility, efficiency, and equity of global health efforts.3 Second, COVID-19 has already questioned and reworked the tools of global health. Resolutions, special sessions, high-level commissions, reports, frameworks, and global action plans can still be useful, but too often global health practice is delivered using the same means as were being used many decades ago. COVID-19, with its limitations on travel, has shown that a strong physical presence within countries is more important than ever.4 Institutions dependent on fly-in, fly-out missions need to rethink their operations. At the same time, COVID-19 has also shown how much of global health travel, especially for generic stakeholder meetings, can be avoided, with benefits for staff, countries, and the planet. Third, COVID-19 has brought to the boil the already simmering discontent about who gets to make decisions in global health and their relationship to global health's intended beneficiaries.5, 6 The discourse on the need to decolonise global health has become prominent and persuasive (although still itself dominated by those based in high-income countries), and the struggle for greater gender equality in its practice and leadership has made important gains. Moreover, the poor performance of many high-resource contexts, including in preventing inequities in outcomes of COVID-19, has made prominent the shifting poles of where public health excellence actually occurs and questioned why global health leadership continues to be dominated by and concentrated in a handful of countries.7 The HIV movement has already shown what is possible in terms of participation and ownership by communities, albeit in unique circumstances. COVID-19 has just re-emphasised lessons that were thought to have been learnt during the west Africa Ebola virus disease outbreak on the importance of community leadership. Fourth, COVID-19 has again shown the limitations of global and national governance in stewarding multisectoral action to tackle complex problems, notwithstanding prominent national exceptions. The construction of false conflicts between public and economic health in response measures has proved disastrous. The siloed nature of health and development policy making and assistance needs urgent attention; global health practice must no longer ignore the essential truth that health is mostly created and destroyed outside of the health sector. Finally and most importantly, the dismal performance in terms of equity for COVID-19's impacts demands a reckoning. Whether in terms of the distribution of commodities such as vaccines, the sadly predictable concentration of mortality in disadvantaged groups within countries, the social and economic impacts of non-pharmaceutical interventions, or access to health care, COVID-19 has shown that, despite decades of rhetoric on the importance of health equity, little has been achieved in terms of mainstreaming its priority within approaches to health. It has never been clearer that attention to the social determinants of health is neither utopian nor abstract but instead fundamental to the effectiveness of public health practice. None of these deficiencies need be terminal. COVID-19 has made the case more persuasively than ever for an effective global health. It is time to use the political oxygen of the current prominence of global health to construct a more participatory, just, and effective practice out of the cruelty and misery of COVID-19 (panel ).Panel Improving the effectiveness and equity of global health practice in the wake of COVID-19 Construct a global health practice that is much more customised and driven by the individual needs of countries and specific populations within countries, elevating decision makers and communities in low-income and lower-middle-income countries to priority actors and audience in global health over global elites. Reboot the global health toolkit, strengthen the focus on in-country presence, and take advantage of the possibilities of remote cooperation, enabling smarter use of virtual and in-person interaction. Centre the rights and perspectives of communities intended as beneficiaries of global health, evaluate the impact of global health in these terms, and accelerate the democratisation and representativeness of global health leadership (including recognising the importance of class and societal position, rather than just diversifying elites). Genuinely transform governance of global health and development to be fit to act across sectors (to prepare better for future infectious disease outbreaks) and revitalise the vision of the Sustainable Development Goals by paying as much attention to required actions outside of the health sector as those within it. Integrate a social determinants approach into pandemic preparedness and global health security efforts, with a proactive focus on equity and identification and prioritisation of marginalised groups between and within countries. The views in this Correspondence are mine alone and do not necessarily reflect the views, policies, or decisions of any of the institutions I have been associated with. I declare no competing interests. ==== Refs References 1 Healthier Societies for Healthy Populations Group Healthier societies for healthy populations Lancet 395 2020 1747 1749 32505243 2 Horton R Offline: The pretensions of global health elites Lancet 395 2020 672 32113492 3 Rajkotia Y Beware of the success cartel: a plea for rational progress in global health BMJ Glob Health 3 2018 e001197 4 Nordström A Is WHO ready to improve its country work? Lancet 390 2017 2749 2751 29303714 5 Abimbola S Pai M Will global health survive its decolonisation? Lancet 396 2020 1627 1628 33220735 6 Rasanathan K Rasanathan JJK Reimagining global health as the sharing of power BMJ Glob Health 5 2020 e002462 7 Dalglish SL COVID-19 gives the lie to global health expertise Lancet 395 2020 1189

33894830

PMC9754103

NO-CC CODE

2022-12-16 23:26:27

no

Lancet. 2021 Apr 22 24-30 April; 397(10284):1543-1544

utf-8

Lancet

10.1016/S0140-6736(21)00713-3

oa_other

==== Front Lancet Lancet Lancet (London, England) 0140-6736 1474-547X Elsevier Ltd. S0140-6736(21)00462-1 10.1016/S0140-6736(21)00462-1 Correspondence Call for a pan-European COVID-19 response must be comprehensive – Authors' reply Priesemann Viola a Brinkmann Melanie M b Ciesek Sandra c Cuschieri Sarah d Czypionka Thomas ef Giordano Giulia g Hanson Claudia hi Hens Niel jk Iftekhar Emil am Klimek Peter no Kretzschmar Mirjam m Peichl Andreas p Perc Matjaž q Sannino Francesco rs Schernhammer Eva l Schmidt Alexander ta Staines Anthony u Szczurek Ewa v a Max Planck Institute for Dynamics and Self- Organization, 37077 Göttingen, Germany b Technische Universität Braunschweig, Helmholtz Zentrum für Infektionsforschung, Braunschweig, Germany c University Hospital, Goethe-University Frankfurt, Frankfurt, Germany d Faculty of Medicine & Surgery, University of Malta, Msida, Malta e Institute for Advanced Studies, Vienna, Austria f London School of Economics and Political Science, London, UK g University of Trento, Trento, Italy h London School of Hygiene & Tropical Medicine, London, UK i Karolinska Institute, Stockholm, Sweden j I-BioStat, Data Science Institute, Hasselt University, Hasselt, Belgium k Centre for Health Economic Research and Modelling Infectious Diseases, Vaccine and Infectious Disease Institute, University of Antwerp, Antwerp, Belgium l Department of Epidemiology, Center for Public Health, Medical University of Vienna, Vienna, Austria m University Medical Center Utrecht, Utrecht, Netherlands n Medical University of Vienna, Vienna, Austria o Complexity Science Hub Vienna, Vienna, Austria p ifo Institute, Leibniz Institute for Economic Research, University of Munich, Munich, Germany q University of Maribor, Maribor, Slovenia r Federico II University of Napoli, Napoli, Italy s Centre of Excellence for Particle Physics and Cosmology and Danish Institute for Advanced Study, University of Southern Denmark, Aarhus, Denmark t Campus Institute for Dynamics of Biological Networks, Göttingen, Germany u School of Nursing, Psychotherapy and Community Health, Dublin City University, Dublin, Ireland v Faculty of Mathematics, Informatics and Mechanics, University of Warsaw, Warsaw, Poland 22 4 2021 24-30 April 2021 22 4 2021 397 10284 15411541 © 2021 Elsevier Ltd. All rights reserved. 2021 Elsevier Ltd Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. ==== Body pmcWe thank Carsten Krüger for the opportunity to further elaborate on the important aspects raised in his Correspondence. First, we are definitely concerned about the impact of lockdowns and continued severe restrictions on the various aspects of human life mentioned. That is precisely why we are committed to the goal of pursuing low case numbers.1 These are necessary to ease restrictions moderately without jeopardising millions of lives and livelihoods.2 The argument is straightforward: uncontained spread would lead to high numbers of deaths and many cases of so-called long COVID, not to mention overburdened health-care systems with collateral impact for everyone. The central idea supporting strong lockdowns and thus rapid case reduction is that strong and effective lockdowns minimise the duration of negative social impacts. In a follow-up statement with coauthors from the humanities and social sciences, we go into more depth on these important socioeconomic aspects.3 Second, we welcome all efforts towards international cooperation and coordination on this matter. Indeed, with this initiative, we hope to promote more cooperation rather than narrow national perspectives. The common legal and institutional framework of the EU allows for a comprehensive joint strategy that can be implemented immediately. A global strategy follows more easily from this important first step. The statement further points to challenges specific to Europe, such as open borders and highly interconnected societies and economies. However, considering that settings of countries all around the world might display larger differences, all these arguments can be immediately generalised to a more global perspective. Vaccinations play a central role in pandemic response. Nonetheless, it will be a while until vaccination takes full effect, and hence we need a sustainable strategy for the coming months and years.3 The vaccination effect will not be sustainable until all people worldwide have had access to vaccination, further underscoringthe importance of international cooperation. Further factors that diminish or slow vaccination progress include obstructions in vaccine production and delivery, the possibility of transmission despite vaccination, and the emergence of escape variants that bypass immunity.3 The latter might even require further vaccine development and the restarting of vaccination programmes. In light of all this, it is necessary that restrictions continue to be maintained for the time being. Hence, low case numbers in the coming months or possibly years, at least below ten per million people per day, have clear benefits for all—for public health, society, and the economy.3 © 2021 Benoit Doppagne/Getty Images 2021 SCi reports grants and personal fees from Roche, and personal fees from Euroimmun, unrelated to this Correspondence. NH reports grants from GSK Biologicals, Pfizer, Merck, and Johnson & Johnson, unrelated to this Correspondence. All other authors declare no competing interests. ==== Refs References 1 Priesemann V Brinkmann MM Ciesek S Calling for pan-European commitment for rapid and sustained reduction in SARS-CoV-2 infections Lancet 397 2021 92 93 33347811 2 Contreras S Dehning J Loidolt M The challenges of containing SARS-CoV-2 via test-trace-and-isolate Nat Commun 12 2021 378 33452267 3 Priesemann V Balling R Brinkmann MM An action plan for pan-European defence against new SARS-CoV-2 variants Lancet 397 2021 469 470 33485462

33894827

PMC9754104

NO-CC CODE

2022-12-16 23:26:27

no

Lancet. 2021 Apr 22 24-30 April; 397(10284):1541

utf-8

Lancet

10.1016/S0140-6736(21)00462-1

oa_other

==== Front Lancet Lancet Lancet (London, England) 0140-6736 1474-547X Elsevier Ltd. S0140-6736(21)00901-6 10.1016/S0140-6736(21)00901-6 World Report Statelessness in the COVID-19 pandemic Burki Talha 22 4 2021 24-30 April 2021 22 4 2021 397 10284 15291530 © 2021 Elsevier Ltd. All rights reserved. 2021 Elsevier Ltd Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. ==== Body pmcMillions of people have no nationality and are being overlooked in the response to COVID-19. Talha Burki reports. There are several ways that you can become stateless. You could have your nationality revoked. Perhaps your father has disappeared, and you live in a place which does not permit mothers to pass on their nationality. Perhaps your country of origin has broken up or been otherwise transformed. Thousands of residents of Kuwait were not granted citizenship after the Gulf state gained independence from the UK in 1961. After the dissolution of the Soviet Union, Estonia and Latvia demanded that ethnic Russians living within their borders pass a language test to become citizens. The two Baltic republics are now home to roughly half of Europe's estimated 528 000 stateless people. But for the most part, statelessness is inherited. If your parents have no nationality, it is extremely hard for you to acquire one. According to the UN High Commissioner for Refugees (UNHCR), 4·2 million people across 76 countries are known to be stateless. However, this figure is likely to represent no more than a third of the overall stateless population. Three-quarters of people without a nationality are from minority groups. The Rohingya were stripped of Burmese citizenship in 1982. The 500 000–600 000 who remain in Myanmar's Rakhine State are subject to discriminatory laws and sporadic pogroms. 2017 saw a brutal crackdown by the Burmese army. Reports emerged of murder, arson, mass rape, and torture. The violence prompted several hundred thousand Rohingya to flee to Bangladesh. The country now hosts almost 1 million Rohingya, 600 000 of whom live in the 23 settlements that make up the Kutupalong-Balukhali Expansion Site. Whether settled or part of refugee or migrant communities, to be stateless is to be vulnerable. Rohingya who attempt to leave Myanmar face an astonishing array of risks from people traffickers, ranging from debt bondage and extortion to slavery. Roma communities in Europe are subjected to xenophobia and hate speech. The invisibility of stateless populations can result in children being excluded from immunisation campaigns, and the lack of documentation makes it difficult to find jobs. If stateless populations have any access to health care, it is usually limited to emergency care. If their residency status is unsettled they risk lengthy periods of detention, and in times of crisis, their vulnerability only increases. A new report from the European Network on Statelessness (ENS) has outlined how Europe's stateless populations have fared during the COVID-19 pandemic. The authors heard from stakeholders in 20 countries. Respondents described the difficulties inherent in maintaining physical distancing in overcrowded settlements and in apartments accommodating several families. An interviewee in Bulgaria noted that the pandemic has been expensive. “You pay to visit the doctor, you pay for the PCR test, you pay for medicine. We have to pay for masks, gloves”, they said. The social security schemes and aid packages that European governments rolled out in the wake of the pandemic have usually been restricted to citizens. “Stateless populations are deprived of a lot of welfare support that others are entitled to”, explains Nina Murray, head of policy and research at the ENS and coauthor of the new report. “They have struggled to mitigate the spread of the virus with non-pharmaceutical interventions and they have to keep working, if they can, which makes it even more difficult to keep their community safe from infection.” Stateless people tend to be in the informal economy, in sectors where pay is contingent on work. For example, they could work as day labourers or sell scrap metal. “A lot of people have suddenly lost their sources of work. They have been pushed into extreme poverty”, Murray told The Lancet. An interviewee for the ENS report recounted the story of a man who had been refused cancer surgery by the UK National Health Service before the pandemic. He was instructed to return to his country of origin for treatment, although the country in question had twice refused to accept him. The report added that there have been instances in which people with severe COVID-19 have died after being unwilling or unable to obtain medical care in the UK. “The British Government has guaranteed that no-one will be charged for COVID-19 care, but there is still a great deal of concern among people with insecure immigration status that they will be billed if they access health care”, said Murray. A survey by R2P, a Ukrainian non-governmental organisation, found that 46% of stateless respondents had been refused access to a community doctor over the course of the pandemic because they lacked identification. Non-citizens often worry that if they make themselves visible to health services, they will be reported to immigration services. The Irish Government has taken steps to ensure that information on patients does not leave the health department, but such measures are rare. Amal de Chickera is co-director of the Institute on Statelessness and Inclusion. He says that the pandemic has put a great deal of pressure on populations that were already over-stretched. In 2019, 1·9 million residents of the Indian state of Assam were left out of the National Register of Citizens. “They have been told they are not citizens, because they cannot prove, through a process that is both arbitrary and discriminatory, that they have been in Assam since before 1971”, explains de Chickera. “They have expended all their resources trying to appeal the decision, getting more and more into debt, and then COVID-19 arrives, and they are not eligible for state relief because they are now viewed as being non-citizens.” By contrast, Sudan has distributed food to vulnerable families during the pandemic without requiring identification. Rohingya refugees off the coast of Indonesia © 2021 Maskur Has/SOPA Images/LightRocket/Getty Images 2021 The atmosphere in Malaysia appears to have become more hostile towards immigrants and refugees. A photograph of a sign outside a mosque in Johor state telling Rohingya they were not welcome went viral. At the end of February, 2021, more than 1000 Rohingya were deported from Malaysia to Myanmar. “There is a real fear among refugees and migrants in Malaysia that if they try to access COVID-related care, they will be detained”, said de Chickera. Many countries shifted services online during the pandemic, which has been problematic for stateless people without stable internet connections. 25 of 45 European countries have confirmed plans to vaccinate stateless populations against COVID-19. They will have to overcome vaccine hesitancy in some communities, including among Roma, who have good reason to distrust the authorities. Countries with small numbers of stateless people might simply vaccinate them along with the general population. However, some nations are actively discriminating against non-citizens. The president of the Dominican Republic has said that those without documents will be excluded from the COVID-19 vaccination campaign, resulting in concerns for the 210 000 residents of Haitian origin who had their nationality rescinded in 2013. An incomplete campaign could be costly, particularly as the Dominican Republic has reported by far the most cases of COVID-19 of any nation in the Caribbean. Most countries with sizeable stateless populations are unlikely to be able to start mass vaccination campaigns this year. Furthermore, the number of stateless people living in the country might be unknown. Issuing information on the vaccines in languages that minority groups can understand might not be a priority. “There is a very real risk that a lot of stateless people will not get the vaccine”, de Chickera told The Lancet. UNHCR is working to ensure that this does not happen. “There are two key considerations”, explains Ann Burton, Chief of the Public Health Section at UNHCR. “The first is whether or not stateless populations are mentioned in national plans. The second is whether they are able to get vaccinated, even if they are mentioned.” Countries that require a national identification number for those who wish to obtain a vaccination could issue temporary numbers to irregular populations. The COVAX Facility aims to ensure that all participating nations have enough vaccines to cover 20% of their population. For cases in which supplies are limited, many governments might not prioritise non-citizens. Gavi, the Vaccine Alliance, has set up a humanitarian buffer, to be deployed in places which have not included populations of concern in national vaccine plans. “The buffer is designed as a last resort, to ensure access to COVID-19 vaccines in settings when there are unavoidable gaps; for example, if the plans do not allocate vaccines to displaced people, including refugees, and asylum seekers”, explains Burton. “It could potentially be used for stateless populations as well.” The diverse nature of statelessness means interventions need to be tailored. Ethnic Russians in Latvia are deprived of political rights, but they are well documented, with full access to health care. They are unlikely to be missed by public services. The same cannot be said of uninsured Roma living in ramshackle settlements, or stateless people who are part of undocumented migrant communities. Murray stresses the importance of extending visas and residency permits during periods of emergency, such as pandemics. “If you regularise stays as, for example, Portugal has done, then you remove a lot of the barriers for accessing health care, state aid, and vaccination drives”, she said. “It is a straightforward measure and it would make a big difference.” Talha Burki was on the expert advisory group for the European Network for Statelessness report but was not involved in researching or writing it.

33894821

PMC9754105

NO-CC CODE

2022-12-16 23:26:27

no

Lancet. 2021 Apr 22 24-30 April; 397(10284):1529-1530

utf-8

Lancet

10.1016/S0140-6736(21)00901-6

oa_other

==== Front J Bus Res J Bus Res Journal of Business Research 0148-2963 0148-2963 Elsevier Inc. S0148-2963(21)00741-4 10.1016/j.jbusres.2021.10.015 Article Remedying Airbnb COVID-19 disruption through tourism clusters and community resilience☆ Jang Seongsoo a⁎ Kim Jinwon b a Cardiff Business School, Cardiff University, Aberconway Building, Colum Drive, Cardiff CF10 3EU, United Kingdom b College of Health and Human Performance, Department of Tourism, Hospitality & Event Management, University of Florida, 186A Florida Gym, PO Box 118208, Gainesville, FL 32611-8208, USA ⁎ Corresponding author. 16 10 2021 2 2022 16 10 2021 139 529542 11 8 2020 2 10 2021 7 10 2021 © 2021 Elsevier Inc. All rights reserved. 2021 Elsevier Inc. Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. Peer-to-peer (P2P) accommodation markets have been disrupted by the COVID-19 pandemic. However, little attention is paid to how to remedy the disruption in terms of P2P accommodation performance. This study empirically investigates the spatially heterogeneous COVID-19 disruptions in the Airbnb business and offers place-based remedying strategies through local resources, including tourism clusters and community resilience. Using real data on Airbnb operating performance and local resources in Florida, we employ spatial econometric models and visualization techniques to estimate the pandemic-disrupted Airbnb performance model. The results show that leisure and hospitality clusters and three resilience resources—social, community capital, and environmental—had spatially heterogeneous effects on Airbnb revenue and booking performance across Floridian counties during the pandemic. Furthermore, community resilience moderated the effect of tourism clusters on Airbnb performance across individual and subclustered counties. These findings enable P2P accommodation hosts and policymakers to adopt destination-specific remedying strategies to cope with the pandemic. Keywords Airbnb COVID-19 Tourism clusters Community resilience Geographically weighted regression ==== Body pmc1 Introduction The coronavirus (COVID-19) pandemic led to a 5.2 percent contraction in global GDP in 2020, although governments put extraordinary efforts into countering the downturn through fiscal and monetary policies (World Bank, 2021). The most disruptive hospitality player, Airbnb, was hit hard by this crisis. Airbnb projected that its revenue in 2020 would decline approximately 54% to $2.2 billion because of the global pandemic (Reuters, 2020). Given lockdown restrictions, Airbnb users have shifted from international travelers to entirely domestic travelers; previously, international Airbnb users constituted 80–90% of all Airbnb users in France, the Netherlands, and Denmark (Chadwick, 2020). Researchers have found that the tourism industry, especially international tourism demand, is vulnerable to external crises or disasters, such as political instability, economic conditions, and natural hazards (Okumus, Altinay, & Arasli, 2005). Recently, researchers have explored the pandemic’s impact on peer-to-peer (P2P) accommodation markets, mainly from the supply perspective. For example, Farmaki et al. (2020) found that host perceptions and responses to the pandemic are categorized into five types in a continuum of optimistic pessimist hosts. Zhang, Geng, Huang, and Ren (2021) also identified three types of postpandemic hosts: innovating entrepreneurs, unchanged diplomats, and quitting speculators. Furthermore, Xu, Huang, and Chen (2021) examined hosts’ health and well-being by exploring their stress and coping strategies after the pandemic. From the demand perspective, Bresciani et al. (2021) investigated how the pandemic and the need for physical distance influence travelers’ choices of different types of accommodation (i.e., Airbnb full flats vs. hotel rooms). Jang, Kim, Kim, and Kim (2021) examined how the interplay between tourists and destination attributes affects P2P accommodation consumption during the pandemic. However, more evidence needs to be applied to the topic of the P2P accommodation demand that has been disrupted by the pandemic. During the ongoing pandemic situation, a focus on the preliminary stage is essential to help P2P accommodation providers and local governments take short-term remedying actions during this crisis and build long-term localized resource development planning in advance of future crises. Compared with other traditional accommodation markets, P2P accommodation markets have emerged as a value cocreation platform to connect tourists with local and authentic experiences at a destination (Guttentag, 2015). Prior research has found that close proximity to leisure and hospitality suppliers might positively (Lee, Jang, & Kim, 2020) or negatively (Zervas, Proserpio, & Byers, 2017) influence P2P accommodation businesses. The colocation of complementary tourism businesses—so-called tourism clusters (Michael, 2003)—creates an overall tourism experience (Gutiérrez, García-Palomares, Romanillos, & Salas-Olmedo, 2017). Furthermore, in the face of a disaster or crisis, destinations need to not only develop economic resources (e.g., tourism businesses) but also reduce resource inequities and address their social vulnerabilities—so-called community resilience (Norris, Stevens, Pfefferbaum, Wyche, & Pfefferbaum, 2008). However, existing research on P2P accommodation disrupted by the pandemic has mainly focused on the role of a destination’s or community’s tangible resources, such as tourism clusters (Jang et al., 2021), and did not incorporate the impact of community resilience as a strategy for destination sustainability and recovery (Hall, 2017, Lin et al., 2017). Hence, of paramount importance is assessing the role of a destination’s immaterial (e.g., social responsibility) and material resources to better understand the pandemic’s impact on P2P accommodation demand (e.g., Hassan & Soliman, 2021). To fill these gaps, this study attempts to empirically examine the spatially heterogeneous effects of local resources on P2P accommodation performance during COVID-19 and offer destination-specific remedying strategies for P2P accommodation businesses that need to deal with the current pandemic. Specifically, this research investigates how two types of local resources—tourism clusters (i.e., leisure and hospitality) and community resilience (i.e., social, community capital, and environmental)—play independent and combined roles in attenuating pandemic-induced P2P accommodation disruption across destinations. As an empirical setting, we selected the U.S. state of Florida because it has widespread COVID-19 transmission, explosive Airbnb growth in rural areas (Florida Trend, 2018), and an array of natural disasters (FDEM, 2020), which have formed different levels of tourism specialization and resilience frameworks across Floridian counties. The results show that subcategories of tourism clusters and community resilience had independent and joint effects on Airbnb’s performance across individual and subclustered counties during the pandemic. This research contributes to a better understanding of the P2P accommodation market that is disrupted by or resilient to the pandemic by incorporating localized material (tourism clusters) and immaterial (community resilience) resources in the performance models. First, this study examines the spatially varying positive and negative effects of tourism clusters and community resilience on P2P accommodation performance during the pandemic. Hence, the findings of this study extend existing P2P accommodation research that primarily deals with hosts’ responses to the pandemic and considers the role of material resources in pandemic-induced disruption. Second, this study identifies the interactive perspective of material and immaterial resources, which better explains how P2P accommodation businesses in some destinations are more resilient to the pandemic than those in other destinations. This finding advances the literature on destination resilience to disasters and crises based on the perspective of the sustainable livelihoods framework for tourism (SLFT), which includes core livelihood assets (e.g., human, economic, social, and natural capital), tourism- and nontourism-related activities and the vulnerability context (Shen, Hughey, & Simmons, 2008). Finally, spatial analytical methods allow P2P accommodation businesses and policymakers to understand the interplay among spatially referenced tourism clusters, resilience, and urban–rural destinations when understanding the complexity of COVID-19-induced P2P accommodation consumption. 2 Literature review 2.1 COVID-19 and P2P accommodation consumption Tourists’ decision making on P2P accommodation use during COVID-19 is likely to be complex from both individual and situational perspectives (Jang et al., 2021, Karl et al., 2020). Pandemic-induced perceived risk is likely to be a significant predictor of tourists avoiding a certain destination and using P2P accommodations. When tourists perceive that potential risks are larger than benefits, they may modify their trip to the destination. Past studies have shown that people normally avoid a trip to places with the spread of a viral infection to reduce their risk of acquiring the disease (e.g., Lau, Griffiths, Choi, & Tsui, 2009). In contrast, despite the virus, some domestic tourists may prepare health and safety procedures in conjunction with their trip (Reisinger & Mavondo, 2005) and voluntarily implement personal nonpharmaceutical intervention (NPI) measures to mitigate their perceptions of risk (Lee et al., 2012). Although perceived travel risk may negatively affect tourists’ decisions to travel to destinations, it is unknown whether tourists, especially domestic tourists, intend to travel to a certain destination and further consume P2P accommodations during COVID-19. From a situational perspective, tourists’ consumption of P2P accommodations tends to be influenced by destination characteristics, such as material attributes (e.g., tourism clusters, Jang et al., 2021) and immaterial attributes (e.g., destination social responsibility, Hassan & Soliman, 2021). Extant studies have identified that the key advantage of P2P accommodations is the authentic local experience from interacting with hosts and other locals at the destination where tourists are staying and traveling (e.g., Mody, Suess, & Lehto, 2017). In addition to accommodation experience, which is central to tourists’ overall destination memorability (Tukamushaba, Xiao, & Ladkin, 2016), the perception of destination localness is a source of authentication of tourists’ consumption experience (Mkono, 2013). Hence, situational factors that P2P accommodation hosts cannot control but that may influence consumers’ overall experiences need to be incorporated as antecedent or moderating variables into P2P accommodation consumption models (Mody et al., 2017, Walls et al., 2011). To date, little empirical research has explicitly examined the role of a destination’s situational factors in shaping tourists’ P2P accommodation demand during COVID-19. 2.2 Roles of tourism clusters and community resilience According to the SLFT, a sustainable tourism livelihoods system includes local assets (e.g., human, social, and natural capital), tourism-related activities (e.g., tourism businesses), institutional arrangements (e.g., local governments), and vulnerability contexts (e.g., shocks) (Shen et al., 2008). A livelihood consists of the capabilities, assets, and activities for making a living to enable sustainable livelihoods to cope with and recover from disasters and crises and maintain or enhance local resources without undermining the natural resource base (Chambers, 1992). The SLFT inherently reveals the multisectoral character of real-life, requiring the integration of both material (e.g., tourism business activities) and immaterial (e.g., social capital) resources into a holistic crisis management framework (Tao & Wall, 2009). In this study, we apply the SLFT perspective to the context of P2P accommodation markets disrupted by COVID-19 and shed light on the role of tourism clusters (i.e., material resources) and community resilience (i.e., immaterial resources) in attenuating the negative impact of the pandemic on P2P accommodation demand. Tourism clusters, which are defined as the specialization of tourism businesses within a particular destination, are crucial for the P2P accommodation business because they provide P2P accommodation consumers with localized tourism experiences (Chan et al., 2012, Lee et al., 2020). Research has found that Airbnb hosts offer limited services and must rely on other tourism products and services that can be served by a number of different firms (Gutiérrez et al., 2017). Tourism clusters can be classified into two categories of tourism industries: leisure (e.g., marinas and golf courses) and hospitality (e.g., hotels and restaurants) businesses (Lee et al., 2020). Furthermore, agglomeration researchers have suggested that the clustering of tourism businesses may increase benefits to members with each additional firm in a cluster—economies of agglomeration—or decrease benefits to members with each additional firm, mainly because a limit is surpassed—diseconomies of agglomeration (McCann & Folta, 2009). Potential sources of agglomeration economies include labor productivity, and those of agglomeration diseconomies include congestion costs (Kim, Williams, Park, & Chen, 2021). For instance, a high clustering of tourist attractions in a particular destination added value to the tourist experience before COVID-19 but was devoid of tourists when the pandemic outbreak occurred in March 2020 (Newman, 2020). Hence, it is imperative to provide empirical evidence on whether and how the effect of tourism clusters on P2P accommodation consumption during the pandemic is positive (i.e., agglomeration economies) or negative (i.e., agglomeration diseconomies) across industries (i.e., leisure and hospitality). Community resilience is defined as “a process linking a network of adaptive capacities (resources with dynamic attributes) to adaptation after a disturbance or adversity” (Norris et al., 2008, p. 127). Community, as an entity with shared geographic boundaries and fate, is composed of “built, natural, social, and economic environments that influence one another in complex ways” (Norris et al., 2008, p. 128). In the face of disasters and crises, individuals experience personal loss, and a community at large shares damages and disruptions to their various environments (Norris, Phifer, & Kaniasty, 1994). During COVID-19, tourists likely perceive resilient destinations as mutually beneficial for tourists and residents because resilience enhances the well-being of locals and tourists’ experiences and offers tourists a safe environment and supportive trip experiences (Hassan & Soliman, 2021). Because SLFT suggests human, social, and natural capital as local livelihood assets (Shen et al., 2008), this study employs three resilience categories—social, community capital, and environmental—as immaterial local resources, whereas tourism clusters are regarded as tourism-related activities (i.e., material resources). Specifically, social resilience captures the demographic qualities of a community’s population (e.g., physical and mental wellness), community capital resilience refers to the goodwill of local citizens to assist their neighbors and fellow citizens during emergencies, and environmental resilience relates to qualities of the environment that enhance the absorptive capacity of natural disasters (Cutter, Ash, & Emrich, 2014). This research attempts to investigate the category of community resilience that plays a critical role in attenuating the negative pandemic effect on P2P accommodation performance. Although abundant research has studied tourism clusters and resilience management, researchers have primarily investigated two dimensions separately without capturing their intersectional effect in traditional and P2P accommodation markets. Regional science researchers argue that the configuration of material and immaterial resources can vary across communities. For example, greater income inequality may attract more skilled and specialized workers in urban U.S. counties but may weaken social cohesion, further hampering agglomeration economies (Fallah & Partridge, 2007). In addition, the concentration of economic activity can be associated with increasing social inequality and can further lead to congestion diseconomies that outweigh agglomeration benefits (Castells-Quintana & Royuela, 2014). Hence, during COVID-19, the concentration of leisure or hospitality businesses in a particular destination may lead to congestion diseconomies that further discourage tourists from traveling to the destination because of a virus infection. Conversely, if the destination is well equipped with high levels of social and environmental resilience, tourists may take their personal NPI measures, travel to this destination, and consume P2P accommodations. In this respect, this study attempts to examine the combined effect of tourism clusters and community resilience on P2P accommodation performance during COVID-19. 2.3 Place-based model for P2P accommodation performance Scholars have agreed that an authentic local experience, as the key contributor to tourists’ overall experience during P2P accommodation stays, includes social interactions with hosts, local residents, and communities (Cheng, 2016, Guttentag, 2015). P2P accommodation services have shifted their travel pattern by offering authentic social experiences to the local community (Cheng, 2016). The important role of the community has also been pronounced during the COVID-19 crisis because each community requires collective, unified action, such as social distancing. Thus, P2P accommodation research needs to go beyond focusing on individual (microlevel) psychological perceptions of the pandemic or state (macro)-level disruptions and examine community (meso)-level disruptions using objective data (Peters, 2020). In this study, we incorporate the destination and community terms in our empirical study and use the concept of the destination community—the location at which tourists spend their time and money and influence the development or degradation of the local environment (Singh, Timothy, & Dowling, 2003). Notably, tourism is closely linked to the social capital and well-being of destination communities (Moscardo, Konovalov, Murphy, & McGehee, 2013). Given the uneven geography of the tourism industry (Lee et al., 2020) and resilience (Cutter, Ash, & Emrich, 2016) resources, we attempt to use spatial analytical methods to identify spatially heterogeneous effects of local resources on pandemic-induced P2P accommodation performance across destination communities. In the accommodation-sharing economy, a destination’s tourism business structure and environment support the growth of tourism-related activities and Airbnb listings in that destination (Gutiérrez et al., 2017). Recently, it has been found that, although the clustering of hospitality businesses (e.g., hotels and restaurants) has a positive impact on Airbnb performance, the relationship between tourism clusters and Airbnb performance has spatial variations across Floridian counties (Lee et al., 2020). In addition to COVID-19 infections, actions to cope with the virus may disproportionately impact communities (Evelyn, 2020). More resilient communities are noted as being less vulnerable to disasters and crises than less resilient communities (Cutter et al., 2014). Specifically, overall resilience in urban areas is primarily driven by economic capital, whereas resilience in rural areas is influenced by social capital with considerable spatial variability (Cutter et al., 2016). In addition, resilience resources (e.g., human and social capital) are configured differently across urban, suburban, and rural areas (Dominiak, 2020). Recent studies on P2P accommodations have found geographical location to be a key attribute for P2P accommodation use (Cheng & Jin, 2019) because it offers tourists unique local experiences. Local experiences during P2P accommodation stays are more important in rural than urban areas, whereas locational benefits in terms of being close to shops and restaurants are deterrent in both urban and rural areas (Mahadevan, 2020). These arguments indicate that a one-size-fits-all approach is not appropriate for boosting Airbnb businesses because of the multidimensional nature of community configurations (Cutter et al., 2016). Entrepreneurial researchers argue that combining community focus with neighboring communities (i.e., translocal embeddedness) to access new ideas and resources is important for entrepreneurial resourcefulness (Kloosterman, 2010). That is, P2P accommodation microentrepreneurs are likely to recognize and create unique opportunities and combine diverse resources by drawing on their embeddedness in the material and immaterial resources of other neighboring communities (Vlasov, Bonnedahl, & Vincze, 2018). To explore the spatially heterogeneous effects of local resources on P2P accommodation performance during COVID-19, we capture two types of spatial heterogeneity: individual and subcluster levels (e.g., Jang, Kim, & von Zedtwitz, 2017). Specifically, the effect of local resources on P2P accommodation performance varies across each destination community. Furthermore, such individual effects may form subclusters because of the translocal embeddedness of local resources and P2P accommodations’ entrepreneurial practices across neighboring local communities (Vlasov et al., 2018). Fig. 1 presents our research model that investigates the independent and joint effects of tourism clusters and community resilience on P2P accommodation performance during COVID-19 within and across communities.Fig. 1 Research model. 3 Methods 3.1 Study area and variables We chose the state of Florida as the empirical study area for several reasons. First, Florida, as one of the most popular tourism destinations, has shown rapid growth in Airbnb development and performance. More than 60,000 Airbnb listings in Florida received $1.2 billion in rent from 6.6 million guests in 2019, reflecting high growth relative to figures for 2018 (45,000 listings, $0.81 billion in rent, 4.5 million guests) and 2017 (40,000 listings, $0.45 billion in rent, 2.7 million guests). Second, the growth rate of Airbnb guests in rural Florida counties has nearly doubled beyond the growth rate in urban counties, indicating that an increasing number of Airbnb users intend to experience rural tourist attractions and not just urban destinations (Florida Trend, 2018). Third, Florida has an array of natural disasters (e.g., sea level rise, hurricane, flooding) that regularly affect local residents and visitors (FDEM, 2020). Finally, on March 1, 2020, Florida became the 7th U.S. state with a documented COVID-19 case and, on April 1, 2020, joined the list of states that limited their residents’ movements and personal interactions outside the home. Such a natural setting enabled us to examine how tourism clusters and community resilience played a critical role in attenuating Airbnb disruptions across urban and rural areas during the early stage of COVID-19 (i.e., March 2020). As the destination community and the unit of analysis, this study employed Floridian County (N = 67) because county-level data are often used for measuring tourism clusters (Lee et al., 2020), community resilience (Cutter et al., 2008), COVID-19 infections (CDC, 2021), and destination-level P2P accommodation performance (Jang et al., 2021). To measure the year-over-year operating performance of Airbnb listings, revenue and booking data for the two months (i.e., March 2019 vs. March 2020) were used in the empirical model because they are commonly used in Airbnb research (Lee et al., 2020, Yang and Mao, 2020). The focus of this study was on how the COVID-19 outbreak affected the growth rate of Airbnb’s operating performance in March 2020 relative to March 2019. Because three datasets (COVID-19, tourism clusters, and community resilience) were collected on a county basis, performance data of individual Airbnb listings acquired from AirDNA were merged at the county level. Finally, the year-over-year growth rates of the average Airbnb revenue-per-available-listing (RevPAL) and average Airbnb occupancy rate (OCR) for each county were defined as the dependent variables. Regarding tourism clusters, two fields—leisure and hospitality—were considered to examine any independent and/or cooperative roles of the leisure and hospitality fields across destination communities (Hobson & Teaff, 1994). The former represents attraction-related businesses, and the latter represents service-related businesses (Lee et al., 2020). To measure the degree of specialization for a specific leisure or hospitality industry in a destination community, the location quotient (LQ) was used because it represents the relative agglomeration of the tourism industry in a county in relation to the entire population (Lazzeretti & Capone, 2006). The LQ can be specified as in Equation (1):(1) LQij=SijStj where sij is the share of tourism industry i’s number of employees in county j relative to the total number of employees in tourism industry i, and stj is the share of county j’s number of employees relative to the total number of employees in the overall U.S. economy. The North American Industry Classification System (NAICS) classifies the arts, entertainment, and recreation industries as NAICS 71 and accommodation and food services as NAICS 72. Whereas NAICS 71 belongs to the leisure industry, NAICS 72 belongs to the hospitality industry (Lee et al., 2020). Finally, the leisure and hospitality LQs for March 2020 were collected and used in the model. To measure three community resilience categories, the Baseline Resilience Indicators for Communities (BRIC) index was used because other resilience measurements focus mainly on place-specific (e.g., urban or rural) or dimension-specific (e.g., infrastructure sector) approaches (Cutter & Derakhshan, 2020). However, the BRIC measurement regards a community as an integrated system that influences crisis/disaster recovery and that consists of six different capitals: social, economic, community capital, institutional, housing/infrastructural, and environmental (Cutter et al., 2014). Each subresilience (e.g., social, community capital, environmental) index is scaled from 0 to 1, with 1 (0) meaning the highest (lowest) resilience among counties in that category. Once constructed, the overall BRIC score can be drawn from summing up six subindex scores, theoretically ranging from 0 to 6 for each county. Finally, the most recent BRIC indexes measured in 2015 were used as the variable of community resilience (Hazards & Vulnerability Research Institute, 2019). Although the 2015 data are not matched with the other variables’ period (2020), the BRIC indexes in Florida showed relatively high stability during the 5-year period (from 2010 to 2015) (Cutter & Derakhshan, 2020). This study controlled three variables that may influence current Airbnb performance. First, Airbnb density—the number of Airbnb listings for a given county—has been found to have a positive agglomeration effect on individual Airbnb listings (Xie, Kwok, & Heo, 2020). Whether the agglomeration effect can play a critical role in attenuating the negative effect of COVID-19 on Airbnb performance is worth identifying. Second, because transportation accessibility influences accommodation prices (Kim, Jang, Kang, & Kim, 2020), this study controlled the effect of the distance to the nearest airport from the county centroid (i.e., airport distance) on Airbnb performance. Finally, the inclusion of population density can control for the effect of the resident population on Airbnb performance because areas with a high population density likely have a wide virus spread, which may further decrease Airbnb consumption. Table 1 presents the operational definitions of all variables used in the model.Table 1 Operationalization of variables and data sources. Variable Operational definition Source Airbnb RevPAL growth Year-over-year percentage change of average Airbnb RevPAL for each county AirDNA Airbnb OCR growth Year-over-year percentage change of average Airbnb occupancy rate for each county Leisure Location quotient of leisure industries (NAICS 71: Arts, Entertainment, and Recreation) for each county US Bureau of Labor Statistics Hospitality Location quotient of hospitality industries (NACIS 72: Accommodation and Food Services) for each county Social Social resilience index for each county Hazards & Vulnerability Research Institute Community capital Community capital resilience index for each county Environmental Environmental resilience index for each county Airbnb density Number (in thousands) of Airbnb listings for each county AirDNA Airport distance Distance (in miles) to the nearest airport from the county centroid Florida Geographic Data Library Population density Number (in thousands) of population for each county US Department of Labor Note: RevPAL: Revenue per available listing; OCR: Occupancy rate; NAICS: North American Industry Classification System. 3.2 Data analysis Multiple data analyses were conducted to measure both the aspatial and spatial effects among variables. First, we ran an ordinary least squares (OLS) regression to examine the global relationships among variables, as shown in Equation (2):(2) yi=β0+∑j=1kβjxj+ε where yi is the dependent variable that consists of Airbnb RevPAL growth (i.e., the growth rate of average Airbnb revenue) and Airbnb OCR growth (i.e., the growth rate of average Airbnb occupancy) in county i∈1,2,⋯,n ; xj is the jth explanatory variable; j ∈1,2,⋯,k ; βj is the jth parameter estimate; and ε is the error term. However, using spatially referenced county-level variables in OLS regression models might lead to biased estimation results from the spatial autocorrelation among variables (Lee, Kim, & Jang, 2021). Thus, a spatial Durbin model (SDM) was also employed to address this issue. The SDM is specified as follows:(3) yi=∑j=1k(ρiWyi+βijxij+Wxijθij)+ui+εi where for an observation in county i, W is the spatial weight, which describes the spatial arrangement for n counties, ρi and θij are the spatial parameters, ui is spatial specific effects, and εi denotes the error term. Next, a geographically weighted regression (GWR) was employed using the same set of variables to explore spatially heterogeneous relationships among variables. Unlike OLS and SDM methods, GWR explores the spatial variation in the relationships between georeferenced variables (Fotheringham, Brunsdon, & Charlton, 2000). GWR has been used as an explorative tool to detect spatial variability over the study area in tourism and hospitality research (Kim et al., 2020, Lee et al., 2020, Lee et al., 2021, Xu et al., 2019). The GWR model is shown in Eq. (4):(4) yi=β0ui,vi+∑j=1kβijui,vixij+εi where (ui, vi) refers to the coordinate at county i’s centroid. The choice of bandwidth is critical for the spatial weighting function. The Gaussian kernel with a fixed bandwidth and a bisquare kernel with adaptive bandwidth are commonly used in GWR. The Gaussian kernel with a fixed bandwidth is suitable when the sample points are regularly distributed in the study area. If the sample points are not regularly spaced, the bisquare kernel with adaptive bandwidth is desirable to accommodate this irregularity. We used a bisquare kernel function because of the geographically different size of county units based on previous regional studies (Lee et al., 2020, Xu et al., 2019). Furthermore, the GWR model fit was maximized when employing the bisquare kernel function compared with the Gaussian kernel function. The optimal kernel size is defined through an iterative optimization approach to minimize the corrected Akaike information criterion (AICc) (Fotheringham, Charlton, & Brunsdon, 1998). Finally, we mapped local GWR coefficients and local R2 to visualize the spatially heterogeneous effects of COVID-19, leisure clusters, hospitality clusters, community resilience, and other control variables on Airbnb performance growth. To analyze the spatial data, we employed advanced software programs, such as ArcGIS Pro, Stata (version 16.1), and GWR (version 4.09). 4 Results 4.1 Descriptive statistics Table 2 reports the descriptive statistics and correlation coefficients for the variables used in the model. Fig. 2 visualizes the spatial distribution of these variables. In Florida, the average year-over-year growth rate of Airbnb RevPAL per county in March 2020 was 0.083, ranging from −0.612 to 3.080 and that of Airbnb OCR per county was −0.017 (mean), ranging from −0.359 to 2.281. Although the average Airbnb performance was positive, Airbnb listings in most (blue-colored) counties suffered negative growth rates (Fig. 2), meaning that Airbnb businesses in Florida were badly disrupted during the early stage of the COVID-19 outbreak. Concerning tourism clusters, the average leisure and hospitality LQs of Florida counties in March 2020 were 1.055 (from 0 to 6.300) and 1.164 (from 0 to 3.660), indicating that county-based tourism industry specialization in Florida was slightly higher than the U.S. county-level average (LQ: 1.0). Regarding community resilience, the average levels of social and community capital resilience in Florida were 0.626 and 0.306, respectively, which are lower than the U.S. average (0.665 and 0.365), but the average level of Floridian environment resilience was 0.636, which is higher than the U.S. average (0.578). Finally, correlation coefficients among independent variables were lower than 0.6, and the highest variance inflation factor (VIF) was 5.013 (Hospitality × Environmental), indicating the absence of multicollinearity in the final model.Table 2 Descriptive statistics and correlation coefficients of variables. Variable Mean Min Max SD (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (1) Airbnb RevPAL growth 0.083 −0.612 3.080 0.504 1.000 (2) Airbnb OCR growth −0.017 −0.359 2.281 0.380 0.649** 1.000 (3) Leisure 1.055 0.000 6.300 1.017 −0.267* −0.301* 1.000 (4) Hospitality 1.164 0.000 3.660 0.641 −0.286* −0.349** 0.477** 1 (5) Social 0.626 0.510 0.704 0.041 −0.401** −0.410** 0.215 0.322** 1 (6) Community capital 0.306 0.191 0.375 0.038 0.008 0.154 −0.350** −0.223 0.203 1 (7) Environmental 0.636 0.550 0.771 0.034 −0.012 −0.002 0.071 0.262* −0.229 −0.195 1 (8) Airbnb density 6.433 0.006 79.447 12.657 −0.226 −0.240 0.274* 0.353** 0.275* −0.486** 0.119 1 (9) Airport distance 30.632 3.130 70.400 16.729 0.035 0.197 −0.342** −0.167 −0.573** 0.007 0.243* −0.234 1 (10) Population density 0.337 0.010 3.034 0.499 −0.201 −0.186 0.346** 0.106 0.336** −0.355** −0.245* 0.358** −0.512** 1 Note: According to Hazards & Vulnerability Research Institute, the US average scores of Social, Community capital, and Environmental are 0.665, 0.365, and 0.578. * p < 0.05; ** p < 0.01. Fig. 2 Spatial distribution of dependent and independent variables used in the model. 4.2 Global model estimations Table 3, Table 4 present the results of two types of global models (i.e., OLS regression model and SDM) using two dependent variables (i.e., Airbnb RevPAL growth and Airbnb OCR growth). The results of Model 1 reveal that, among the two types of tourism clusters, leisure clusters had a negative effect on Airbnb revenue performance (β= −0.165, p < 0.05), whereas hospitality clusters had no effect. This finding implies that a stronger dependence on leisure-related businesses and their employment caused more serious damage to Airbnb revenue during the COVID-19 crisis. Interestingly, social resilience had a negative effect on Airbnb revenue (β= −5.217, p < 0.05) for Model 1, whereas community capital resilience had a positive effect on Airbnb booking (β= 1.202, p < 0.05) for Model 4. This finding indicates the differential role of resilience resources in the P2P accommodation business during the pandemic. Furthermore, the results show that social resilience played a crucial role in attenuating the negative effect of hospitality clusters on Airbnb revenue marginally (Model 1: β= 1.375, p < 0.10) and on Airbnb bookings significantly (Model 4: β= 7.986, p < 0.05).Table 3 Estimation of OLS, SDM, and GWR models (DV: Airbnb RevPAL growth). Variable OLS (Model 1) SDM (Model 2) GWR (Model 3) Min Mean Max DIFF Spatial weight Queen contiguity Kernel function using adaptive bi-square Leisure −0.165** −0.089** −46.437 11.094 94.800 −6.621 Hospitality −0.001 −0.021 −2.012 −0.070 5.638 −18.041 Social −5.217** −1.908** −12.133 −0.515 2.532 −9.749 Community capital −1.787 −6.402 −43.981 −0.156 48.426 −8.333 Environmental 0.061 1.548 −111.094 6.081 122.407 −15.291 Leisure × Social 2.574 6.100 −117.691 −17.561 20.143 −7.308 Leisure × Community capital 6.755 7.208 −28.320 17.185 97.570 −7.082 Leisure × Environmental −0.097 −1.423 −202.166 −1.244 120.433 −8.710 Hospitality × Social 1.375* 4.354* −28.336 11.378 69.658 −8.255 Hospitality × Community capital 1.481 −11.595 −84.336 −7.459 41.366 −7.855 Hospitality × Environmental 0.773 3.884 −127.211 2.977 122.663 −2.559 Airbnb density −0.007 −0.015 −152.590 −4.761 124.096 −3.002 Airport distance −0.014** −0.012** −0.438 −0.014 0.243 −9.633 Population density −0.111 −0.046 −0.102 −0.024 0.018 −7.634 W × Leisure 0.134 W × Hospitality −0.271 W × Social 0.983 W × Community capital −20.166 W × Environmental 9.843 W × Leisure × Social −7.160 W × Leisure × Community capital 14.715** W × Leisure × Environmental −26.783 W × Hospitality × Social 13.172 W × Hospitality × Community capital –33.184 W × Hospitality × Environmental −0.010 W × Airbnb density −0.046 W × Airport distance −0.032** W × Population density −1.102** Intercept 4.470 5.231 −46.437 11.094 94.800 R2 0.471 0.483 0.485 0.501 0.526 ρ 0.001 σ2 0.070** Note: DIFF denotes difference of criterian value. * p < 0.10; ** p < 0.05. Table 4 Estimation of OLS, SDM, and GWR models (DV: Airbnb OCR growth). Variable OLS (Model 4) SDM (Model 5) GWR (Model 6) Min Mean Max DIFF Spatial weight Queen contiguity Kernel function using adaptive bi-square Leisure −0.084 −0.126 −3.464 0.128 4.495 −6.081 Hospitality −0.002 −0.031 −18.067 −1.780 1.083 −18.013 Social −3.569 −0.898 −30.185 1.729 82.702 −9.786 Community capital 1.202** 0.683** −105.391 39.624 651.792 −8.299 Environmental 0.940 0.201 −200.033 −14.707 20.025 −15.286 Leisure × Social 0.457 1.036 −52.805 8.786 120.529 −7.246 Leisure × Community capital −1.035 0.404 −243.564 16.911 517.541 −6.707 Leisure × Environmental 1.130 −1.510 −208.038 −9.260 36.813 −8.732 Hospitality × Social 9.509** 7.986** −21.064 5.182 61.926 −8.907 Hospitality × Community capital 0.468 5.417 −46.832 39.736 509.813 −7.859 Hospitality × Environmental −0.084 −0.519 −84.375 −4.735 38.897 −2.597 Airbnb density −0.005 −0.010 −0.316 0.037 1.058 −3.132 Airport distance −0.002 −0.002 −0.060 −0.002 0.101 −9.612 Population density 0.154 0.066 −54.819 −4.054 1.446 −7.765 W × Leisure 0.010 W × Hospitality 0.201 W × Social 1.772 W × Community capital −15.970** W × Environmental 4.958** W × Leisure × Social −7.239 W × Leisure × Community capital 9.652** W × Leisure × Environmental −10.453* W × Hospitality × Social 8.658 W × Hospitality × Community capital −28.773** W × Hospitality × Environmental −2.808 W × Airbnb density −0.043** W × Airport distance −0.013** W × Population density −0.248 Intercept 1.305 0.957 −111.869 −2.309 38.053 R2 0.549 0.575 0.521 0.594 0.657 ρ 0.001 σ2 0.033** Note: DIFF denotes difference of criterian value. * p < 0.10; ** p < 0.05. The SDM results (Model 2 in Table 3 and Model 5 in Table 4) also confirmed the results of the OLS regression model. For example, leisure clusters had a negative effect on Airbnb revenue performance (β= −0.089, p < 0.05) for Model 2, whereas hospitality clusters also had no effect. In addition, social resilience had a negative effect on Airbnb revenue (β= −1.908, p < 0.05) for Model 2, whereas community capital resilience had a positive effect on Airbnb booking (β= 0.683, p < 0.05) for Model 5. This finding indicates significant effects of tourism clusters and community resilience on Airbnb performance across destination communities. Due to the significant spatial spillover effects of W*variables on Airbnb performance, the SDM estimated spatial feedback loop influences (Lesage and Fischer 2008), which identify the average effect of the independent variables on Airbnb performance of a county compared to its neighboring counties and vice versa (Kim et al., 2021). For example, the coefficient of the spatial spillover of Leisure × Community capital for Model 2 is 14.715, indicating a positive spatial spillover effect (Table 3). However, because these results do not reflect the marginal effects of X on Y, the estimations of the direct, indirect, and total effect of each variable on Airbnb performance can infer more accurate interpretations of the spillover effects (see the details in Appendix A). 4.3 Local model estimations To examine the existence of spatial variability among local coefficients, we estimated the difference of criterion (DIFF) value, which identifies the difference of AICc between the fitted GWR and a model with the k-th coefficient fixed and all other coefficients kept as they are in the fitted GWR (Latinopoulos, 2018, Nakaya, 2015). If the DIFF value is greater than 2, the corresponding variable has no significant spatial variability and could be better predicted by a global term in the model. The last column of Table 3 reports that the DIFF values were below −2, revealing significant spatial variation in all local coefficients across Floridian counties. Specifically, Model 3 reports that leisure clusters, on average, positively affect Airbnb RevPAL growth (βMean = 11.094); however, depending on the county, the relationship was negative (βMin = -46.437) or more positive (βMax = 94.800). A similar pattern existed for the Airbnb OCR growth model (Model 6), for which local coefficients ranged from −3.464 to 4.495 (βMean = 0.128). To provide a better understanding of the spatially varying coefficients, Fig. 3, Fig. 4 map the spatial distribution of local GWR coefficients for eight variables of tourism clusters and their interactions with community resilience across counties in the Airbnb RevPAL growth model and the Airbnb OCR growth model, respectively. For example, leisure clusters had a negative effect on the operating performance of Airbnb businesses in some northwest Floridian (blue-colored) counties but a positive effect on those in other northwest (red-colored) counties. In addition, hospitality clusters had a positive effect on the revenue performance of Airbnb listings in some northcentral Floridian (red-colored) counties but a negative effect in southeast (blue-colored) counties. These findings reveal that the relationship between tourism clusters and Airbnb business performance varied across counties, and tourism clusters played a critical role in enhancing the performance of Airbnb listings, especially in rural and less populated counties. Similarly, three categories of community resilience had mixed (positive or negative) effects on Airbnb performance across Floridian counties.Fig. 3 Spatial distribution of local GWR coefficients in Airbnb RevPAL growth model. Fig. 4 Spatial distribution of local GWR coefficients in Airbnb OCR growth model. Interestingly, the combined effects of both tourism clusters and community resilience on Airbnb performance varied depending on variable combinations and locations (Fig. 3). From the perspective of rural counties, the negative effect of leisure clusters on Airbnb revenue was attenuated by high levels of social resilience in northwest counties and environmental resilience in southwest counties. In urban counties, community capital resilience attenuated the negative effect of leisure clusters on Airbnb revenue in northwest and central counties, and environmental resilience played a critical role in reducing the negative effect of hospitality clusters on Airbnb revenue in south counties. These results indicate that, from the perspective of Airbnb performance, some communities that rely heavily on leisure or hospitality clusters were vulnerable to external shocks, such as the COVID-19 pandemic, but might reduce this disruption with greater community resilience. In addition, compared with the OLS regression, the GWR improved the overall explanatory power of the Airbnb performance model (i.e., R2 in Table 3, Table 4), and the two Airbnb performance models performed better in northwest Floridian (dark-colored) counties (Fig. 3, Fig. 4). The results of GWR estimations imply that the effect of tourism clusters and community resilience on Airbnb performance varies across individual destination communities. Finally, based on the obtained local coefficients, an application study to explore the subclustering of high or low local coefficients was performed using the global Moran’s I statistic and local indicators of spatial association (LISA). In this study, the global Moran’s I measures whether spatial dependence exists among the county-level coefficient of a focal location and coefficients of other neighboring locations (Li, Calder, & Cressie, 2007). LISA cluster maps can be classified into 5 types of spatial clusters: (1) high-high: hot spots; (2) high-low: spatial outliers; (3) low–high: spatial outliers; (4) low-low: cold spots; and (5) not significant (Jang & Kim, 2018; Jang et al., 2017). Fig. 5, Fig. 6 illustrate the spatial distribution of hot and cold spots across variables. For example, Airbnb listings in the red-colored cluster of low- and mid-populated northwest counties benefitted from the combination of leisure clusters and social resilience that led to better revenue performance during COVID-19 (Fig. 5). In contrast, Airbnb listings in the cluster of densely populated South Floridian counties benefitted from the association of hospitality clusters with community capital resilience (Hospitality × Community capital: red-colored) but not from their association with social resilience (Hospitality × Social: blue-colored). These results show the existence of spatial heterogeneity at the subcluster level in terms of the role of local resources in shaping P2P accommodation performance during COVID-19.Fig. 5 Spatial distribution of clustered GWR-based local coefficients in Airbnb RevPAL growth model. Fig. 6 Spatial distribution of clustered GWR-based local coefficients in Airbnb OCR growth model. 5 Discussion and conclusion COVID-19 has heavily hit the tourism and lodging industry, especially small tourism businesses such as Airbnb listings that are vulnerable to crises and disasters. Using combined data on Airbnb performance and local resources in 67 Floridian counties over March 2019 and 2020, this study used spatial econometric models and GIS techniques and further examined the spatially heterogeneous effects of leisure clusters, hospitality clusters, and resilience resources (i.e., social, community capital, and environmental) on the growth rates of Airbnb and booking performance. The results of global regression models show that leisure clusters and social resilience negatively influenced Airbnb revenue, community capital resilience positively influenced Airbnb bookings, and social resilience attenuated the negative effect of hospitality clusters on both Airbnb revenue and bookings. In addition, the results of local regression models indicate that Airbnb listings in rural counties with a high specialization of leisure and hospitality businesses were less disrupted by COVID-19 than those in urban counties. Furthermore, although community resilience had mixed effects on Airbnb performance across counties, it moderated the spatially varying relationship between tourism clusters and Airbnb performance. For example, social (community capital) resilience attenuated the negative effect of leisure clusters on Airbnb revenue in some rural (urban) counties, whereas environmental resilience attenuated the negative effect of leisure (hospitality) clusters on Airbnb revenue in some rural (urban) counties. Such positive and negative relationships were heterogeneous across individual and subclustered counties, implying the existence of spatial heterogeneity in the Airbnb performance model. 5.1 Theoretical implications Our study contributes to the literature on P2P accommodation and tourism crisis management by empirically investigating the spatially heterogeneous effects of destination-specific situational factors on P2P accommodation performance during COVID-19. First, our study reveals the importance of tourism clusters and community resilience to understand spatially heterogeneous P2P accommodation business disruptions and prepare for a future pandemic crisis (Jang et al., 2021). This finding implies that authentic local experience, as a key advantage for P2P accommodations (Mody et al., 2017), is embedded with tourism clusters and community resilience that accommodation hosts cannot control (Walls et al., 2011) during a disaster or crisis. This study extends the concept of SLFT to the context of the accommodation-sharing economy by identifying material (i.e., leisure and hospitality clusters) and immaterial (i.e., social, community capital, and environmental resilience) resources as sustainable livelihoods for P2P accommodation businesses during the pandemic crisis (Tao & Wall, 2009). In addition, exploring two situational factors that might influence tourists’ P2P accommodation consumption during COVID-19 contributes to the P2P accommodation literature because recent studies have separately examined their effects (Hassan and Soliman, 2021, Jang et al., 2021). Second, this study offers evidence for the spatially heterogeneous effects of tourism clusters on Airbnb performance during COVID-19, thereby extending tourism cluster theory (Michael, 2003). The empirical findings identified the existence of both economies and diseconomies of tourism business agglomeration across urban and rural destinations in terms of P2P accommodation performance. Prior studies have mainly focused on the positive role of tourism clusters (i.e., agglomeration economies) in P2P accommodation markets (Gutiérrez et al., 2017, Lee et al., 2020), which did not consider the pandemic context. This study filled this gap by showing that agglomeration economies of leisure businesses enhanced Airbnb revenue in both rural and urban destinations (Fig. 5) but Airbnb bookings mainly in rural destinations (Fig. 6). Such mixed results can be explained by the complexity in tourists’ decision making during the pandemic (Karl et al., 2020). Business tourists might intend to use Airbnb listings in urban destinations during the pandemic, whereas leisure tourists likely traveled to less populated rural destinations with leisure attractions (Jang et al., 2021). Conversely, agglomeration diseconomies of hospitality businesses disrupted Airbnb performance mainly in urban destinations (e.g., Miami-Dade County) to which business and leisure tourists often travel. The reason may be that a high clustering of hotels and restaurants in urban destinations was likely to increase congestion costs (McCann & Folta, 2009), spread the virus easily during the pandemic, and in turn discourage tourists from traveling to the destinations. This finding extends to the literature on agglomeration economies and diseconomies in the context of tourism clusters and P2P accommodation markets. Finally, the combined effects of tourism clusters and community resilience resonate with research showing the importance of sustainable livelihoods in the form of material and immaterial resources across destination communities (Shen et al., 2008). Previous research has identified that overall community resilience is driven by different resilience resources, such as economic capital for urban resilience and community capital for rural resilience (Cutter et al., 2016). This study advances the resilience literature by showing the heterogeneous roles of specific resilience resources in P2P accommodation markets during the pandemic. Interestingly, the combination of leisure clusters with social resilience increased Airbnb performance in rural and less populated urban destinations, whereas the association of hospitality clusters with community capital resilience enhanced Airbnb performance in more populated urban destinations. This finding confirms that P2P accommodation consumers with greater social resilience, such as physical and mental wellness, are likely to travel to rural destinations than to urban destinations (Mahadevan, 2020, Mody et al., 2017). Furthermore, suburban areas often combine the features of rural communities, such as a higher level of trust (Dominiak, 2020) and community capital (e.g., local citizens’ goodwill to assist their neighbors), and can remedy the pandemic-induced diseconomies of hospitality business agglomeration for P2P accommodations in urban destinations. 5.2 Practical implications Our empirical findings offer practical insights for stakeholders in the P2P accommodation economy. Specifically, this research suggests that Airbnb listings need to take full advantage of two types of situational factors (i.e., tourism clusters and community resilience) in their counties to attenuate the negative impact of COVID-19 on their revenue and booking performance. Airbnb hosts are encouraged to conduct a detailed analysis of specialized tourism clusters (e.g., leisure and hospitality) and the configuration of community resilience (e.g., social, community capital, environmental) that generate agglomeration (dis)economies and crisis management capacity, respectively, and reflect these components in their during- or post-COVID-19 marketing activities. For example, Airbnb hosts in rural destinations need to utilize the agglomeration economies of leisure resources (e.g., parks and beaches) and specific resilience factor(s) in their destinations in terms of product offerings and communication messages to potential guests. In addition, urban and rural Airbnb hosts could target business and bleisure (business and leisure) tourists, respectively, because tourists will more voluntarily implement personal measures to avoid viral infection on their trips to the destination (Jang et al., 2021). From a policy perspective, local governments should take a place-based approach to manage the current pandemic crisis and prepare for a future crisis because a one-size-fits-all strategy cannot reflect the multidimensional nature of local resource configurations (Cutter et al., 2016). Depending on the geographical features of individual and neighboring communities, policymakers need to understand how the combination of material and immaterial resources attenuates the negative impact of COVID-19 on the Airbnb business across individual and subclustered destination communities. For example, because northwest Florida has rich leisure resources (e.g., historical places, white beaches, national forests, and natural springs), Airbnb hosts could leverage the advantage of outdoor leisure clusters and resilience resources across neighboring counties. According to 2021 Airbnb search data, two northwest Floridian beaches (i.e., Cape San Blas and Grayton Beach) are among the top destinations for Airbnb users because the pandemic makes Airbnb users stay in unique and remote lodgings with plenty of privacy and outdoor space (Hayes, 2021). To maximize P2P accommodation consumption, this study suggests that each county in northwest Florida should be embedded with greater social resilience through a greater concentration of physicians and mental health supporting facilities. In addition, the findings of southeastern Florida (i.e., highly populated urban destinations) suggest that although diseconomies of hospitality agglomeration disrupted Airbnb revenue during COVID-19, greater environmental resilience attenuated the disruptions. For example, Greater Miami, as a resilient city, needs to enhance environmental resilience through urban forests, solar energy initiatives, and other climate change mitigation efforts (Caraway-Carlton, 2019), which can enhance the image of socially responsible destinations with potential tourists and Airbnb users. 5.3 Limitations and future research directions Although the findings are insightful, several study limitations exist. First, because the empirical models are specific to the population of Floridian Airbnb listings, the findings of this study cannot be applied to other regions and countries. Future research can resolve the applicability issue by collecting and analyzing empirical data related to Airbnb performance, tourism clusters, and community resilience from other study areas. Second, this research focused on the early stage of the COVID-19 crisis and failed to capture how the relationship between local resources and Airbnb business performance evolved during the period. Future studies can resolve this limitation by collecting longitudinal data during the pandemic lifecycle. Third, although our SDM results showed the existence of both direct and (indirect) spillover effects of local resources on P2P accommodation performance, this study focused mainly on the spatially varying direct effects across individual and subclustered destinations. Future studies can explore the spatially heterogeneous spillover effects to identify where the competitive and complementary effects of local resources on P2P accommodation performance exist. Finally, due to a multicollinearity issue among variables, this study did not decompose tourism clusters and community resilience into detailed components. For instance, leisure subindustries (e.g., art, entertainment, and recreation) and other community resilience resources, such as economic, housing/infrastructure, and institutional resources, can be included in the model. By using advanced modeling techniques, future studies can use decomposed resource components to explain the specific set of local resources that leads to better Airbnb performance during the pandemic. Declaration of Competing Interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Seongsoo Jang. Seongsoo Jang is an interdisciplinary marketing researcher and a Senior Lecturer of Marketing at Cardiff Business School, Cardiff University, UK. His research interests include digital marketing and spatial analytics in retailing, tourism, and hospitality. Jinwon Kim. Jinwon Kim is a tourism/recreation/community geographer and an Assistant Professor in the Department of Tourism, Hospitality and Event Management at the University of Florida, USA. His research goal is to identify the role of tourism, recreation and park in the creation of active, vibrant, healthy, sustainable and resilient communities. Appendix A Estimations of direct, indirect, and total effects in SDM coefficients Table A1 shows the direct, indirect, and total effects of each variable on Airbnb revenue and booking performance. The direct effect measures the average effect of the change in independent variable (X) on dependent variable (Y), which includes the feedback via the neighboring county and back to the focal county. The indirect effect measures the average effect of the change in X of the focal county on the Y of neighboring counties. The total effect denotes the sum of the direct and indirect effects, which measures the average effect of the change in X of the focal county on the Y of all the focal and neighboring counties. For example, the direct effect of leisure clusters on Airbnb revenue was larger than the coefficient estimate (−0.199 vs. −0.089), implying the existence of feedback effect that passed via neighboring counties back to the focal county. Interestingly, the indirect effect of leisure clusters was positive (0.059) so that the total effect was smaller than the direct effect (-0.141), although both were statistically non-significant. This implies that there might be a decrease in Airbnb user traffic in a focal county with a high degree of clustering of leisure attractions, whereas neighboring counties with relatively lower leisure specialization might benefit from the spillover effect. In addition, both the direct and indirect effects of Hospitality × Social on Airbnb booking were positive (7.289 and 1.101), which lead to the positive total effect (8.390). This indicates that social resilience attenuated the pandemic-induced P2P accommodation market disruption across focal and neighboring counties.Table A1 Impact measures: direct, indirect, and total effects in SDM coefficients. Variable DV (Airbnb RevPAL growth) DV (Airbnb OCR growth) Direct Indirect Total Direct Indirect Total Leisure −0.199** 0.059 −0.141 −0.159 0.050 −0.108 Hospitality −0.008 −0.292 −0.299* −0.017 0.041 0.024 Social −1.695** 0.278 −1.372 0.916 1.576 2.491 Community capital −3.300 −4.234 −7.535** 2.397** −10.698** −8.301** Environmental 3.324 7.117** 10.441** 0.159 3.531* 3.681** Leisure × Social 6.712 −4.893 1.819 2.463 −3.824 −1.361 Leisure × Community capital 4.447 3.468 7.914** −0.251 6.253* 6.002** Leisure × Environmental −2.064 −12.132** −14.196* 0.333 −5.144 −4.811 Hospitality × Social 4.126 3.274 7.400 7.289** 1.101 8.390 Hospitality × Community capital 2.110 −4.189 −2.079 4.007 −12.582** −8.575** Hospitality × Environmental 0.841 −1.300 −0.459 −0.957 −1.069 −2.025 Airbnb density −0.007 −0.006 −0.014 −0.007 −0.020** −0.027** Airport distance −0.013** −0.010** −0.023** 0.001* −0.007* −0.006** Population density −0.119 −0.429** −0.548** 0.079 −0.310 −0.232 ☆ Dr. Jinwon Kim’s participation was supported by the Ministry of Education of the Republic of Korea and the National Research Foundation of Korea (NRF-2019S1A3A2098438). ==== Refs References Bresciani S. Ferraris A. Santoro G. Premazzi K. Quaglia R. Yahiaoui D. Viglia G. The seven lives of Airbnb. The role of accommodation types Annals of Tourism Research 88 2021 103170 Caraway-Carlton, A. (2019). Miami’s environment & sustainability. Greater Miami Convention & Visitors Bureau. (accessed April 14, 2021), [Retrieved from https://www.miamiandbeaches.com/about-gmcvb/miami-s-environment-sustainability]. Castells-Quintana D. Royuela V. Agglomeration, inequality and economic growth The Annals of Regional Science 52 2 2014 343 366 CDC (2021). Centers for Disease Control and Prevention. COVID data tracker. (accessed May 26, 2021), [Retrieved from https://covid.cdc.gov/covid-data-tracker/#county-view]. Chadwick, L. (2020). Coronavirus: Hotels and Airbnb plan ‘fundamental shift’ after COVID-19 lockdowns. (accessed July 01, 2020), [Retrieved from https://www.euronews.com/2020/05/08/coronavirus-hotels-and-airbnb-plan-fundamental-shift-after-covid-19-lockdowns]. Chambers R. Sustainable livelihoods: The poor’s reconciliation of environment and development Ekins P. Max-Neef M. Real-life economics: Understanding wealth creation 1992 Routledge London 214 230 Chan L.-Y. Lin H.-L. Wang C.-L. Industry-region position and economic performance of travel and tourism service industry: An agglomeration perspective Asia Pacific Journal of Tourism Research 17 5 2012 562 576 Cheng M. Sharing economy: A review and agenda for future research International Journal of Hospitality Management 57 2016 60 70 Cheng M. Jin X. What do Airbnb users care about? An analysis of online review comments International Journal of Hospitality Management 76 2019 58 70 Cutter S.L. Derakhshan S. Temporal and spatial change in disaster resilience in US counties, 2010–2015 Environmental Hazards 19 1 2020 10 29 Cutter S.L. Ash K.D. Emrich C.T. The geographies of community disaster resilience Global Environmental Change 29 2014 65 77 Cutter S.L. Ash K.D. Emrich C.T. Urban-rural differences in disaster resilience Annals of the American Association of Geographers 106 6 2016 1236 1252 Cutter S.L. Barnes L. Berry M. Burton C. Evans E. Tate E. Webb J. A place-based model for understanding community resilience to natural disasters Global Environmental Change 18 4 2008 598 606 Dominiak J. Human and social capital in urban agglomeration and its changes in the core-peripheries pattern. The example of Poznań agglomeration City, Culture and Society 23 2020 100371 Evelyn, K. (2020). ‘It’s a racial justice issue’: Black Americans are dying in greater numbers from Covid-19. (accessed July 01, 2021), [Retrieved from https://www.theguardian.com/world/2020/apr/08/its-a-racial-justice-issue-black-americans-are-dying-in-greater-numbers-from-covid-19]. Fallah B.N. Partridge M. The elusive inequality-economic growth relationship: Are there differences between cities and the countryside? The Annals of Regional Science 41 2 2007 375 400 Farmaki A. Miguel C. Drotarova M.H. Aleksić A. Časni A.Č. Efthymiadou F. Impacts of Covid-19 on peer-to-peer accommodation platforms: Host perceptions and responses International Journal of Hospitality Management 91 2020 102663 10.1016/j.ijhm.2020.102663 32901166 FDEM. (2020). Florida Division of Emergency Management. [Retrieved from https://www.floridadisaster.org/]. Florida Trend. (2018). Report: Rural Florida experiencing explosive Airbnb growth. (accessed July 01, 2020), [Retrieved from https://www.floridatrend.com/article/25460/report-rural-florida-experiencing-explosive-airbnb-growth]. Fotheringham S.A. Brunsdon C. Charlton M. Quantitative geography: Perspectives on spatial data analysis 2000 Sage Publications London Fotheringham A.S. Charlton M.E. Brunsdon C. Geographically weighted regression: A natural evolution of the expansion method for spatial data analysis Environment and Planning A 30 11 1998 1905 1927 Gutiérrez J. García-Palomares J.C. Romanillos G. Salas-Olmedo M.H. The eruption of Airbnb in tourist cities: Comparing spatial patterns of hotels and peer-to-peer accommodation in Barcelona Tourism Management 62 2017 278 291 Guttentag D. Airbnb: Disruptive innovation and the rise of an informal tourism accommodation sector Current Issues in Tourism 18 12 2015 1192 1217 Hall C.M. Resilience in tourism: Development, theory, and application Cheer J.M. Lew A.A. Tourism 2017 Resilience and Sustainability. Routledge Hassan S.B. Soliman M. COVID-19 and repeat visitation: Assessing the role of destination social responsibility, destination reputation, holidaymakers’ trust and fear arousal Journal of Destination Marketing & Management 19 2021 100495 10.1016/j.jdmm.2020.100495 Hayes, K. (2021). Two Florida beaches rank as top destinations for Airbnb users this spring. Florida Politics. (accessed April 12, 2021), [Retrieved from https://floridapolitics.com/archives/412713-two-florida-beaches-rank-as-top-destinations-for-airbnb-users-this-spring/]. Hazards & Vulnerability Research Institute. (2019). Baseline Resilience Indicators for Communities (BRIC). BRIC 2015 County Scores. University of South Carolina, https://artsandsciences.sc.edu/geog/hvri/bric. Hobson J.S.P. Teaff J.D. Hospitality and leisure/recreation: Towards an understanding of an emerging partnership serving the tourism industry Journal of Hospitality and Leisure Marketing 2 1 1994 43 54 Jang S. Kim J. Remedying food policy invisibility with spatial intersectionality: A case study in the Detroit Metropolitan Area Journal of Public Policy & Marketing 37 1 2018 167 187 Jang S. Kim J. Kim J. Kim S. Spatial and experimental analysis of peer-to-peer accommodation consumption during COVID-19 Journal of Destination Marketing & Management 20 2021 100563 10.1016/j.jdmm.2021.100563 Jang S. Kim J. von Zedtwitz M. The importance of spatial agglomeration in product innovation: A microgeography perspective Journal of Business Research 78 2017 143 154 Karl M. Muskat B. Ritchie B.W. Which travel risks are more salient for destination choice? An examination of the tourist’s decision-making process Journal of Destination Marketing & Management 18 2020 100487 10.1016/j.jdmm.2020.100487 Kim J. Jang S. Kang S. Kim S.(. Why are hotel room prices different? Exploring spatially varying relationships between room price and hotel attributes Journal of Business Research 107 2020 118 129 Kim Y.R. Williams A.M. Park S. Chen J.L. Spatial spillovers of agglomeration economies and productivity in the tourism industry: The case of the UK Tourism Management 82 2021 104201 10.1016/j.tourman.2020.104201 Kloosterman R.C. Matching opportunities with resources: A framework for analysing (migrant) entrepreneurship from a mixed embeddedness perspective Entrepreneurship & Regional Development 22 1 2010 25 45 Lazzeretti L. Capone F. Identification and analysis of local tourist systems: An application to Italy (1996–2001) Lazzeretti L. Petrillo C.S. Tourism Local Systems and Networking 2006 Elsevier) (Amsterdam-New York 25 42 Latinopoulos D. Using a spatial hedonic analysis to evaluate the effect of sea view on hotel prices Tourism Management 65 2018 87 99 Lau J.T.F. Griffiths S. Choi K.C. Tsui H.Y. Widespread public misconception in the early phase of the H1N1 influenza epidemic Journal of Infection 59 2 2009 122 127 19592114 Lee C.-K. Song H.-J. Bendle L.J. Kim M.-J. Han H. The impact of non-pharmaceutical interventions for 2009 H1N1 influenza on travel intentions: A model of goal-directed behavior Tourism Management 33 1 2012 89 99 32287736 Lee Y.-J. Jang S. Kim J. Tourism clusters and peer-to-peer accommodation Annals of Tourism Research 83 2020 102960 10.1016/j.annals.2020.102960 Lee Y.-J.-A. Kim J. Jang S. Intertemporal tourism clusters and community resilience The Professional Geographer 2021 1 6 Lesage J.P. Fischer M.M. Spatial growth regressions: Model specification, estimation and interpretation Spatial Economic Analysis 3 3 2008 275 304 Li H. Calder C. Cressie N.A. Beyond Moran's I: Testing for spatial dependence based on the spatial autoregressive model Geographical Analysis 39 4 2007 357 375 Lin Y. Kelemen M. Kiyomiya T. The role of community leadership in disaster recovery projects: Tsunami lessons from Japan International Journal of Project Management 35 5 2017 913 924 Mahadevan R. Is there an urban-rural divide in the demand for peer-to-peer accommodation sharing? Current Issues in Tourism 2020 1 8 McCann B.T. Folta T.B. Demand- and supply-side agglomerations: Distinguishing between fundamentally different manifestations of geographic concentration Journal of Management Studies 46 3 2009 362 392 Michael E.J. Tourism micro-clusters Tourism Economics 9 2 2003 133 145 Mkono M. Using net-based ethnography (netnography) to understand the staging and marketing of ‘Authentic African’ dining experiences to tourists at Victoria falls Journal of Hospitality & Tourism Research 37 2 2013 184 198 Mody M.A. Suess C. Lehto X. The accommodation experiencescape: A comparative assessment of hotels and Airbnb International Journal of Contemporary Hospitality Management 29 9 2017 2377 2404 Moscardo G. Konovalov E. Murphy L. McGehee N. Mobilities, community well-being and sustainable tourism Journal of Sustainable Tourism 21 4 2013 532 556 Nakaya, T. (2015). Geographically weighted generalised linear modelling. In Brunsdon, & Singleton (Eds.), Geocomputation: A practical primer. London, UK: Sage Publishing. Newman, J. (2020). The Coronavirus has emptied tourist sites and the images are haunting. (accessed April 13, 2021), [Retrieved from https://www.townandcountrymag.com/leisure/travel-guide/g31483152/empty-cultural-landmarks-coronavirus/]. Norris F.H. Stevens S.P. Pfefferbaum B. Wyche K.F. Pfefferbaum R.L. Community resilience as a metaphor, theory, set of capacities, and strategy for disaster readiness American Journal of Community Psychology 41 2008 127 150 18157631 Norris F. Phifer J. Kaniasty K. Individual and community reactions to the Kentucky floods: Findings from a longitudinal study of older adults Ursano R. McCaughey B. Fullerton C. Individual and community responses to trauma and disaster: The structure of human chaos 1994 Cambridge University Press Cambridge 378 400 Okumus F. Altinay M. Arasli H. The impact of Turkey's economic crisis of February 2001 on the tourism industry in Northern Cyprus Tourism Management 26 1 2005 95 104 Peters D.J. Community susceptibility and resiliency to COVID-19 across the rural-urban continuum in the United States Journal of Rural Health 36 3 2020 446 456 Reisinger Y. Mavondo F. Travel anxiety and intentions to travel internationally: Implications of travel risk perception Journal of Travel Research 43 3 2005 212 225 Reuters (2020). Airbnb warns 2020 revenue can fall by half amid COVID-19 hit: The Information. (accessed July 01, 2020), [Retrieved from https://www.reuters.com/article/us-airbnb-outlook/airbnb-warns-2020-revenue-can-fall-by-half-amid-covid-19-hit-the-information-idUSKCN21Q35D]. Shen F. Hughey K.F.D. Simmons D.G. Connecting the sustainable livelihoods approach and tourism: A review of the literature Journal of Hospitality and Tourism Management 15 01 2008 19 31 Singh S. Timothy D.J. Dowling R.K. Tourism in destination communities 2003 CABI Wallingford Tao T.C.H. Wall G. Tourism as a sustainable livelihood strategy Tourism Management 30 1 2009 90 98 Tukamushaba E. Xiao H. Ladkin A. The effect of tourists’ perceptions of a tourism product on memorable travel experience: Implications for destination branding European Journal of Tourism, Hospitality and Recreation 7 1 2016 2 12 Vlasov M. Bonnedahl K.J. Vincze Z. Entrepreneurship for resilience: Embeddedness in place and in trans-local grassroots networks Journal of Enterprising Communities: People and Places in the Global Economy 12 3 2018 374 394 Walls A.R. Okumus F. Wang Y.(. Kwun D.-W. International journal of hospitality management an epistemological view of consumer experiences International Journal of Hospitality Management 30 1 2011 10 21 World Bank. (2021). Global Economic Prospects. (accessed March 15, 2020), [Retrieved from https://www.worldbank.org/en/publication/global-economic-prospects]. Xie K.L. Kwok L. Heo C.Y. Are neighbors friends or foes? Assessing Airbnb listings’ agglomeration effect in New York City Cornell Hospitality Quarterly 61 2 2020 128 141 Xu X.(. Huang D. Chen Q. Stress and coping among micro-entrepreneurs of peer-to-peer accommodation International Journal of Hospitality Management 97 2021 103009 10.1016/j.ijhm.2021.103009 Xu Y.-H. Pennington-Gray L. Kim J. The sharing economy: A geographically weighted regression approach to examine crime and the shared lodging sector Journal of Travel Research 58 7 2019 1193 1208 Yang Y. Mao Z. Location advantages of lodging properties: A comparison between hotels and Airbnb units in an urban environment Annals of Tourism Research 81 2020 102861 10.1016/j.annals.2020.102861 Zervas G. Proserpio D. Byers J.W. The rise of the sharing economy: Estimating the impact of Airbnb on the hotel industry Journal of Marketing Research 54 5 2017 687 705 Zhang M.o. Geng R. Huang Y. Ren S. Terminator or accelerator? Lessons from the peer-to-peer accommodation hosts in China in responses to COVID-19 International Journal of Hospitality Management 92 2021 102760 10.1016/j.ijhm.2020.102760 33199933

0

PMC9754115

NO-CC CODE

2022-12-16 23:26:27

no

J Bus Res. 2022 Feb 16; 139:529-542

utf-8

J Bus Res

10.1016/j.jbusres.2021.10.015

oa_other

==== Front J Am Med Dir Assoc J Am Med Dir Assoc Journal of the American Medical Directors Association 1525-8610 1538-9375 Published by Elsevier Inc. on behalf of AMDA -- The Society for Post-Acute and Long-Term Care Medicine. S1525-8610(22)00887-8 10.1016/j.jamda.2022.11.009 Original Study - Brief Report Nudge-Based Interventions on Health Promotion Activity Among Very Old People: A Pragmatic, 2-Arm, Participant-Blinded Randomized Controlled Trial Yamada Yukari PhD a Uchida Tomoe MPH a Sasaki Shusaku PhD b Taguri Masataka PhD c Shiose Takayuki PhD d Ikenoue Tatsuyoshi PhD ae Fukuma Shingo PhD a∗ a Fukuma Research Group, Human Health Sciences, Kyoto University Graduate School of Medicine, Sakyo, Kyoto, Japan b Center for Infectious Disease Education and Research, Osaka University, Suita City, Osaka, Japan c Department of Health Data Science, Tokyo Medical University, Tokyo, Japan d Kyoto University Museum, Yoshida Honmachi, Kyoto, Japan e Center for Data Science Education and Research, Shiga University, Shiga, Japan ∗ Address correspondence to Shingo Fukuma, MD, PhD, Fukuma Research Group, Human Health Sciences, Kyoto University Graduate School of Medicine, 54 Shogoin-Kawahara, Sakyo, Kyoto 606-8507, Japan. 15 12 2022 15 12 2022 7 9 2022 4 11 2022 9 11 2022 © 2022 The Authors 2022 Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. Objectives Social distancing due to the coronavirus disease 2019 crisis can exacerbate inactivity in older adults. Novel approaches for older adults must be designed to improve their activity and maintain their health. This study examined the effect of nudge-based behavioral interventions on health-promoting activities in older adults in Japan. Design Two-arm, participant-blinded randomized controlled trial. Setting and Participants Japanese continuing care retirement community residents (n = 99, median age 82 years, 73% women) Intervention Two-step nudge-based behavioral intervention promoting tablet usage. Methods We enrolled participants from an ongoing Internet of Things project in a retirement community in Japan. For the health promotion program, tablet computers were installed in a common area for participants to receive information about their health. The intervention group received a 1-time loss-emphasized nudge (first step), followed by asking questions about when they planned to use it again (second step). The control group used the tablet computers without being asked those questions. The main outcome was the participants’ mean daily tablet activity every 4 weeks for the next 16 weeks. Results Ninety-nine individuals were randomly assigned to the intervention or control group. The rate ratios for tablet use were significantly higher in the intervention group in the second and third periods. The subgroup analysis showed that these effects were largely attributable to men. Conclusions and Implications Nudge-based interventions can be effective in promoting activities for older adults, especially older men. The finding of this study indicates a possible intervention to engage people who are socially isolated. Keywords Nudge activity promotion aged 80 and over RCT COVID pandemic ==== Body pmcThe social distancing policy established because of the coronavirus disease 2019 (COVID-19) reduced physical and social activities worldwide,1 , 2 raising concerns about harmful health consequences, especially for older adults.3, 4, 5 As the damage to health resulting from inactivity increases with age,6 , 7 changing the sedentary lifestyle of older adults is an urgent and important public health issue. Although there is evidence of effective interventions to promote healthy behavior in older adults, such as educational counseling,8, 9, 10 resources are usually needed for each targeted individual for short-term success. Nudge theory, a less costly approach that can lead to behavioral changes, has received widespread attention in research and policy making.11 , 12 Nudge is defined as “an aspect of choice architecture that predictably alters people’s behavior without forbidding any options or significantly changing their economic incentives.”13 Nudge can guide people to personally and socially desirable choices by targeting their subconscious biases and routines that are present in human decision-making processes. In their theoretical framework, humans tend to prefer avoiding losses to acquiring equivalent gains14; thus, behavioral changes could be encouraged by emphasizing losses rather than benefits (“loss-emphasized nudge”).15 Humans also tend to procrastinate on important tasks,16 and declaring their planned execution date could reinforce their intention to change their behaviors17 (“commitment nudge”). Nudge-based interventions have been adopted for a wide range of behaviors, such as chronic disease management, and social distance promotion.13 , 18 , 19 However, few studies have assessed the effect of nudge-based interventions on the very old population (aged ≥80 years); therefore, the usefulness of nudge-based messages in a highly aging society remains unclear. As it is hypothesized that physical and social changes during the aging process can influence the underlying mechanism of how nudges affect people’s behaviors, in this study, we intentionally targeted the very old population and examined whether nudge-based interventions resulted in behavioral changes to promote their levels of activity using a senior-friendly communication device. Methods Study Design This trial was conducted as part of an ongoing research project in a continuing care retirement community (CCRC) in Kyoto, Japan.20 The research project, which started in 2018, aimed to promote the health of older adults through a senior-friendly Internet of Things. It has 30 beacon sensors covering the site and voluntary participants carry a low-energy Bluetooth beacon. Given the infrastructure, older adults do not need additional operations and receive a feedback sheet showing their walking distance, the areas they visited, and the time they spent in the common areas. The project also involved the installation of tablet computers in the common areas intended to guide participants out of their rooms. The tablet computer senses a beacon approaching and starts an individually customized talk flow with audio assistance, allowing older users to navigate easily. The talk flow consists of, but is not limited to, a greeting message according to the extent of daily walking distance, questions about daily health status, and rock-paper-scissors games with the CCRC staff’s photos. They use a touchscreen stylus of either their own or one provided at the site with an alcohol disinfectant (Supplementary Figure 1.). Data derived from beacons and the tablet computers were linked by research ID to participants’ administrative data, including frailty status defined by an annual CCRC self-report survey using the Kihon checklist (Supplementary Table 1).21 Participants provided written consent for participation in the project, and the Institutional Review Board approved the study (R1669). We randomized the project participants on July 1, 2021, and compared daily activities of tablet computer use from July 13, 2021, for 16 weeks between those who received a series of nudge-based messages and those who did not. Study Population We assessed the eligibility of all study participants who carried their beacon card daily based on beacon-detected data during the 3 months prior to randomization (ie, from April 1, 2021, to June 30, 2021). Because habituality of tablet computer usage strongly influences the outcome of interest of this study, we excluded participants whose tablet computer use during the same period was defined as an outlier after transforming to a standard normal distribution.22 Randomization and Masking To address possible contaminations in intervention status caused by cohabitation, we excluded one of the cohabitants from the randomization process if both were study participants. Furthermore, we stratified the participants by their history of tablet computer use (yes/no). They were then separately randomized to either group in a 1:1 ratio using computer-generated random numbers. After randomization, the excluded participants were assigned to the same group as their cohabitant. The first author generated a random allocation sequence and assigned the participants to the intervention. Participants and care providers at the site were blinded to interventions. Intervention The intervention was constructed based on previous studies showing that classic nudges work better when combined with reflective elements23 and that older adults may benefit from game-based interventions.24 Our nudge-based intervention consisted of 2 steps: a loss-emphasized message and a commitment device. The first step was the provision of information on the time-limited special content available on tablet computers (“Time-limited special quizzes made by Kyoto University are available from July 13 for a week! Don’t miss it!”). The message was printed on a regular feedback sheet for the intervention group and posted in their individual CCRC mailboxes on July 12, 2021. The second step was designed to reinforce users’ intentions to promote their activity. When using tablet computers during the week after the first step was administered, only those in the intervention group received a simple question on the tablet computer: “When would you use the tablet computer next time?” They were asked to select one of the options: “today or tomorrow,” “within a week,” or “after one week.” The control group received the regular feedback sheet without the additional message and used the time-limited special quizzes on tablet computers but did not receive the additional question. Supplementary Table 2 summarizes the content delivered to the intervention and control groups. The possibility of participants knowing their intervention status was limited in 2 senses. First, because the regular feedback sheet was in an envelope and posted in their mailbox, they were unlikely to share it with others. Second, tablet computers were expected to be used individually, as the simultaneous presentation of multiple beacons near the tablet computers would cause confusion in detecting individuals. Outcomes and Follow-up The outcome measure was the participants’ mean daily activity using tablet computers every 4 weeks for the following 16 weeks. The daily activity was counted as the number of responses recorded on tablet computers. The responses to the commitment question were not included as they were only recorded in the intervention group. Statistical Methods We conducted an intention-to-treat analysis; all participants were analyzed in the group to which they were allocated, regardless of whether they received the second step of the intervention. First, we drew a time series graph of the rolling average of 3 days of the mean daily use of the tablet computer according to the groups. Second, differences in the means of tablet computer use every 4 weeks between the groups were analyzed as a whole and separately by sexes. Lastly, we regressed daily usage on an interaction term of the intervention status with dummy variables for the elapsed period since the intervention (ie, 0, pre-period; 1, first period, including the intervention week; and 2, second period) to obtain a period-specific differentiated effect of the intervention status as a whole and separately by sexes. A generalized estimating equation model assuming a Poisson distribution with an exchangeable correlation structure was used as panel data clustered by participants with count outcomes (log-link) using a bootstrap estimation. The results of the sensitivity analyses of those living alone and those defined as tablet computer users in the previous 3 months were used to determine any unintended bias caused in the randomization sequence. Furthermore, sex and mean daily count of tablet computer use during the preintervention period were included separately as covariates in the main analysis as a control for the potentially disproportional distribution of participants’ characteristics between the groups. All analyses were performed with Stata, version 15.0 (StataCorp). Results Study Population A total of 110 residents who provided valid informed consent as of June 30, 2021 were selected for eligibility; 99 met the eligibility criteria. All randomized participants completed the trial (Supplementary Figure 2). Table 1 shows participants’ basic characteristics by group. The median age was 82.1 years (interquartile range, 78.1-87.2), and 72 participants (73%) were women. The prevalence of frailty status defined by the annual self-report survey indicated that the study participants were similar in frailty to the corresponding age groups of the general population of the community (Supplementary Table 3).25 Their walking distances inside the CCRC were slightly less than 1 km, and they spent an average of approximately 1 hour in the common areas per day.Table 1 Baseline Characteristics of Participants Intervention (n = 52) Control (n = 47) All (N = 99) Demographics  Median age (Q1, Q3),∗ y 83.8 (78.3, 87.4) 81.6 (78.1, 86.3) 82.1 (78.1, 87.2)  <75 4 (7.7) 5 (10.6) 9 (9.1)  ≥75 and <80 11 (21.2) 11 (23.4) 22 (22.2)  ≥80 and <85 17 (32.7) 18 (38.3) 35 (35.4)  ≥85 20 (38.5) 13 (27.7) 33 (33.3)  Females 33 (63.5) 39 (83.0) 72 (72.7)  No cohabitant 37 (71.2) 39 (83.0) 76 (76.8) Frailty status†  Physical function and strength domain 29 (55.8) 23 (48.9) 52 (52.5)  Malnutrition domain 3 (5.8) 4 (8.5) 7 (7.1)  Oral function and eating 29 (55.8) 26 (55.3) 55 (55.6)  Socialization and housebound domain 14 (26.9) 10 (21.3) 24 (24.2)  Cognitive and memory domain 24 (46.2) 20 (42.6) 44 (44.4)  Depression and mood domain 33 (63.5) 33 (70.2) 66 (66.7) Beacon-derived activities, mean (SD)‡  Daily social relation, h 1.01 (0.81) 1.24 (0.89) 1.12 (0.85)  Daily distance, m 698.89 (418.33) 796.05 (722.17) 745.02 (581.60)  Daily visited spots, n 4.49 (2.76) 4.91 (3.04) 4.69 (2.89) Values are presented as numbers (percentages), unless stated otherwise. ∗ As of June 30, 2021. † Obtained from the Kihon checklist taken in the CCRC in February 2021, developed by the Japanese Ministry of Health, Labour and Welfare as a publicly used screening tool to identify signs of frailty. Detailed items and cutoff values are available in Supplementary Table 1. ‡ Obtained from beacon logs between April 1 and June 30, 2021. Main Outcomes Figure 1 shows the 3-day rolling average of the mean daily activity of the tablet computer. The activities were similar between the groups before the intervention and then increased during the intervention week (gray shade), with more drastic increases observed in the intervention group than in the control group (red line). There were no differences between the groups immediately after the intervention week; however, approximately a month later, some differences were observed between the groups, which continued for 12 weeks. Similar trends were observed in both men and women, with a larger difference observed among men than among women.Fig. 1 Three-day rolling average of daily tablet activity. n = 52 for the intervention group, n = 47 for the control group. Dot lines indicate a 4-week interval. Figure 2 shows the primary outcomes expressed as the differences in the mean daily activities of the tablet computers of the participants every 4 weeks between the groups (intervention minus control). The daily activities of the intervention group were significantly higher in the second and third periods than those of the control group (P = .003 and .047, respectively). Men were more responsive to the intervention than women. Detailed results are provided in Supplementary Table 4.Fig. 2 Differences in daily count of tablet computer usage every 4 weeks between intervention and control group. Bars indicate 95% CIs. Positive value indicates the intervention group used tablet more frequently than the control group. Pre-period: June 15 to July 12, 2021; first period: July 13 to August 9, 2021; second period: August 10 to September 6, 2021; third period: September 7 to October 4, 2021; fourth period: October 5 to November 1, 2021. The rate ratios of the intervention group for daily activities of the tablet computers compared to the preintervention period and the control group also showed that the likelihood of the intervention group using tablet computers during the second and third periods was more than twice the control group (mean rate ratio = 2.38, 95% CI 1.21, 4.71; and 2.16, 95% CI 1.14, 4.08, respectively). A subgroup analysis by sex revealed that the effect of the intervention was mainly in men (Supplementary Table 5). Sensitivity Analyses Sensitivity analyses with only those living alone and those with a history of tablet computer use showed results similar to the main analyses. The models including sex or previous tablet computer activities showed they were not significant covariates and did not change the main results (Supplementary Table 6). Discussion Our results indicate that our 2-step behavioral intervention, which involved loss-emphasized nudge and commitment nudge, made a difference by increasing older adults’ health promotion activities in the following 12 weeks. The findings indicated that nudges can be effective in the short and long term when combined appropriately. Promoting older adults’ activities is the key to designing health services for them. Therefore, the findings of this study will help improve the health outcomes of older adults. The initial large effect of the nudges in the intervention group disappeared quickly; however, the intervention group experienced a subtle but consistent increase in their activities after a few weeks, and the increase persisted for at least 8 weeks. These findings suggest the 2-step nudge-based intervention worked as expected, highlighting that loss of opportunity generates short-term behavioral change through an automatic and affective system that influences immediate behavioral decisions (heuristics)26 and that the provision of commitment devices facilitates the translation of newly evolved behavior into regular behaviors (deliberation).23 Additionally, the findings are partially consistent with an experimental study suggesting the “temporal spillover effect” of nudges, in which aimed behavioral changes were observed subsequently without the presence of a nudge.27 In this study, the underlying mechanism was hypothesized to be that the initial behavioral change mediated the effect of nudges, resulting in a prolonged effect. The magnitude of the effect varied with sex; men were more responsive to the intervention. The exact element of our intervention that prompted older men to promote their activity could not be determined, but our intervention involving technology and affirmation (eg, “Kyoto University–made quizzes”) might have worked better for men than women among older adults.28 Women are more likely than men to engage in social and physical activity, especially in older age29 , 30; thus, our study findings may be novel for indicating a possible intervention to engage socially isolated people. The limitations of this study include sampling bias due to the reliance on participants in an existing project in a CCRC. We chose the CCRC because the Internet infrastructure we created enabled us to examine older adults’ immediate and detailed behaviors in response to the message delivered and to offer high-quality, low-cost, rapid trials even during the COVID-19 pandemic, where safety measures would have prevented us from conducting any trials targeting older adults. Another limitation is the outcome that is not a direct indicator of health behaviors. We assessed nudge-based messages to promote tablet usage and found its effect in this study. Then it is presumable that nudges to promote physical activity promote older adults’ physical activity. Future alternative studies could be conducted using a wearable device to assess the effect of nudge-based interventions directly on physical activity in older adults. Conclusions and Implications This study revealed that nudge-based interventions can be effective in promoting activities for older adults, especially older men. This finding indicates a possible intervention to engage socially isolated people. Uncited Supplementary Figure Supplementary Figure 3 Supplementary Material Supplementary Fig. 1 Enrollment. • Those whose beacons were not sensed between April 1 and June 30, 2021, were excluded from the 110 residents with valid informed consent as of June 30, 2021. • Outliers of tablet use in the previous 3 months were defined using the Grubbs test (Pearson/df = 11.146, with a threshold of α = 2). • If a participant lived with someone who was also eligible, the one with a lower research ID number was removed temporarily at this point. They were assigned to the same group as their cohabitants after randomization. • Those who used the tablet at least once between April 1 and June 30, 2021, were thought to be tablet users. Supplementary Fig. 2 Onsite photos. Photograph of a tablet computer site. Supplementary Fig. 3 A ballpoint pen with a touch rubber tip with the project logo printed on the side. Supplementary Table 1 Detailed Items of Self-Reported Questionnaire (Kihon Checklist) 1 Do you go out by bus or train by yourself? 0. Yes 1. No 2 Do you go shopping to buy daily necessities by yourself? 0. Yes 1. No 3 Do you manage your own deposits and savings at the bank? 0. Yes 1. No 4 Do you sometimes visit your friends? 0. Yes 1. No 5 Do you turn to your family or friends for advice? 0. Yes 1. No 6 Do you normally climb stairs without using handrails or wall for support? 0. Yes 1. No 7 Do you normally stand up from a chair without any aids? 0. Yes 1. No 8 Do you normally walk continuously for 15 minutes? 0. Yes 1. No 9 Have you experienced a fall in the past year? 1. Yes 0. No 10 Do you have a fear of falling while walking? 1. Yes 0. No 11 Have you lost 2 kg or more in the past 6 months? 1. Yes 0. No 12 Height: __ cm, weight: __ kg, BMI: __kg/m2. If BMI is less than 18.5, this item is scored 1. Yes 0. No 13 Do you have any difficulties eating tough foods compared to 6 months ago? 1. Yes 0. No 14 Have you choked on your tea or soup recently? 1. Yes 0. No 15 Do you often experience having a dry month? 1. Yes 0. No 16 Do you go out at least once a week? 0. Yes 1. No 17 Do you go out less frequently compared to last year? 1. Yes 0. No 18 Do your family or your friends point out your memory loss? Eg, "You always ask the same question over and over again"? 1. Yes 0. No 19 Do you make a call by looking up phone numbers? 0. Yes 1. No 20 Do you find yourself not knowing today’s date? 1. Yes 0. No 21 In the last two weeks have you felt lack of fulfilment in your daily life? 1. Yes 0. No 22 In the last two weeks have you felt a lack of joy when doing the things you used to enjoy? 1. Yes 0. No 23 In the last two weeks have you felt difficulty in doing what you could do easily before? 1. Yes 0. No 24 In the last two weeks have you felt helpless? 1. Yes 0. No 25 In the last two weeks have you felt tired without a reason? 1. Yes 0. No Physical function/strength domain is applicable if 3 or more of questions 6-10 were rated 1. Malnutrition domain is applicable if 2 or more of questions 11-12 were rated 1. Oral function and eating domain is applicable if 2 or more of questions 13-15 were rated 1. Socialization and housebound domain is applicable if either 16 or 17 was rated 1. Cognitive and memory domain is applicable if 2 or more of questions 18-20 were rated 1. Depression and mood domain is applicable if 2 or more of questions 21-25 were rated 1. Supplementary Table 2 Contents Delivered to Intervention and Control Groups Intervention Group Control Group Received a routine feedback sheet on 12th July X X Received a loss-framed message printed on the feedback sheet (“Time-limited special Kyoto University–made quizzes are available from July 13 for a week! Don’t miss it!”) X Enjoyed the special quizzes on tablet computers between 13th and 20th July X X Answered to the question “When would you use the tablet computer next time?” on tablet computers between 13th and 20th X Supplementary Table 3 Percentages of Frailty Status: Comparison of the Study Participants With an Available Statistic From the General Community-Dwelling Older Population Sex Comparative General Population∗ Study Population Men Women Age 75-79 80-84 85-89 75-79 80-84 85-89 75-79 80-84 85-89 n 1149 664 254 1368 861 393 22 35 33 Physical function/strength domain 23.4 31.0 44.0 34.2 50.4 61.2 31.8 44.1 72.7 Malnutrition domain 2.1 4.5 7.6 3.5 3.8 5.8 9.0 2.9 9.0 Oral function/eating 21.8 28.0 41.1 26.7 35.4 35.4 31.8 58.8 63.6 Socialization/housebound domain 38.4 46.9 60.9 45.1 59.3 61.9 40.9 25.7 18.2 Cognitive/memory domain 18.1 27.6 40.7 15.5 22.6 29.3 50.0 37.1 51.5 Depression/mood domain 32.5 41.1 50.3 36.9 46.3 46.3 54.5 60.0 78.8 ∗ Adapted from Kameoka study participants who are without the long-term care insurance certificate. Yamada Y, Nanri H, Watanabe Y, et al. Prevalence of frailty assessed by Fried and Kihon checklist indexes in a prospective cohort study: design and demographics of the Kyoto-Kameoka Longitudinal Study. J Am Med Dir Assoc. 2017;18:733.e7–733.e15. Supplementary Table 4 Daily Count of Tablet Computer Use of Groups Intervention Control Mean Difference (95% CI) P Value∗ All n = 1456 person-day n = 1316 person-day  Preintervention period 0.383 (0.040) 0.448 (0.040) −0.066 (−0.178, 0.047) .087  First period 0.635 (0.081) 0.508 (0.055) 0.128 (−0.069, 0.324) .30  Second period 0.376 (0.046) 0.185 (0.025) 0.191 (0.086, 0.296) .003  Third period 0.506 (0.053) 0.275 (0.032) 0.231 (0.107, 0.355) .047  Fourth period 0.445 (0.054) 0.318 (0.031) 0.127 (−0.003, 0.258) .68 Females n = 924 person-day n = 1092 person-day  Preintervention period 0.435 (0.054) 0.422 (0.041) 0.013 (−0.119, 0.1448) .26  First period 0.514 (0.100) 0.523 (0.063) −0.009 (−0.234, 0.217) .005  Second period 0.264 (0.040) 0.198 (0.028) 0.066 (−0.027, 0.160) .50  Third period 0.470 (0.061) 0.266 (0.034) 0.204 (0.073, 0.336) .14  Fourth period 0.445 (0.054) 0.318 (0.031) 0.067 (−0.066, 0.200) .89 Males n = 532 person-day n = 224 person-day  Preintervention period 0.291 (0.058) 0.576 (0.124) −0.285 (−0.520 to −0.049) .16  First period 0.846 (0.137) 0.433 (0.102) 0.413 (–0.022, 0.848) .18  Second period 0.570 (0.104) 0.121 (0.053) 0.449 (0.126, 0.772) .001  Third period 0.570 (0.099) 0.321 (0.088) 0.248 (−0.071, 0.567) .38  Fourth period 0.577 (0.109) 0.393 (0.102) 0.184 (−0.170, 0.539) .77 Values are presented as mean (SE), unless stated otherwise. Preintervention period: June 15 to July 12, 2021; first period: July 13 to August 9, 2021; second period: August 10 to September 6, 2021; third period: September 7 to October 4, 2021; fourth period: October 5 to November 1, 2021. ∗ Two-sample Wilcoxon rank-sum test. Supplementary Table 5 Mean Rate Ratios and 95% CIs of Intervention Status for Daily Tablet Computer Use According to Sex All (N = 99) Females (n = 72) Males (n = 27) First period 1.47 (0.86, 2.50) 0.95 (0.41, 2.20) 3.86 (1.84, 8.09) Second period 2.38 (1.21, 4.71) 1.30 (0.68, 2.47) 9.34 (4.24, 20.59) Third period 2.16 (1.14, 4.08) 1.72 (0.91, 3.25) 3.50 (1.21, 10.15) Fourth period 1.64 (0.76, 3.54) 1.18 (0.55, 2.57) 2.90 (1.20, 7.00) The values are exponentiated coefficients of the interaction term between the intervention status and the period. The references are the preintervention period and control group. First period: July 13 to August 9, 2021; second period: August 10 to September 6, 2021; third period: September 7 to October 4, 2021; fourth period: October 5 to November 1, 2021. Supplementary Table 6 Rate Ratios and 95% CIs of Intervention Status for Daily Tablet Usage in the Sensitivity Analyses Only Those Living Alone (n = 76) Only Those With History of Tablet Use in the Previous 3 mo ahead of the Intervention (n = 56) Adjusted by Sex (N = 99) Adjusted by Pre Tablet Activity (N = 99) First period 1.49 (0.77, 2.87) 1.48 (0.78, 2.80) 1.47 (0.82, 2.62) 1.47 (0.89, 2.42) Second period 2.27 (0.96, 5.38) 2.39 (1.00, 5.69) 2.38 (1.08, 5.66) 2.38 (1.15, 4.94) Third period 1.58 (0.78, 3.17) 2.15 (1.05, 4.38) 2.16 (1.12, 4.16) 2.16 (1.16, 4.00) Fourth period 1.29 (0.49, 3.43) 1.64 (0.71, 3.79) 1.64 (0.72, 3.74) 1.64 (0.83, 3.23) Men 1.71 (0.27, 11.07) Pre tablet activity 1.02 (1.01, 1.03) The values are exponentiated coefficients of the interaction term between the intervention status and the period. References are the preintervention month and the control group. Acknowledgments We thank all staff and residents of Kyoto Yuyunosato and GOCCO Inc for their immeasurable cooperation. This study is supported by the 10.13039/501100001691 Japan Society for the Promotion of Science (grants 17KT0041 and 19H03959). SF has received salary as a part-time physician from the clinic attached to the CCRC since April 2021. TU received a salary as a part-time worker from the CCRC between July 2019 and March 2021. The other authors declare no conflicts of interest. ==== Refs References 1 McCarthy H. Potts H.W.W. Fisher A. Physical activity behavior before, during, and after COVID-19 restrictions: Longitudinal smartphone-tracking study of adults in the United Kingdom J Med Internet Res 23 2021 e23701 33347421 2 Tison G.H. Avram R. Kuhar P. Worldwide effect of COVID-19 on physical activity: A descriptive study Ann Intern Med 173 2020 767 770 32598162 3 Heid A.R. Cartwright F. Wilson-Genderson M. Pruchno R. Challenges experienced by older people during the initial months of the COVID-19 pandemic Gerontologist 61 2021 48 58 32955079 4 Le Couteur D.G. Anderson R.M. Newman A.B. COVID-19 through the lens of gerontology J Gerontol A Biol Sci Med Sci 75 2020 e119 e120 32222763 5 Yamada Y. Uchida T. Ogino M. Ikenoue T. Shiose T. Fukuma S. Changes in older people’s activities during the Coronavirus disease 2019 pandemic in Japan J Am Med Dir Assoc 21 2020 1387 1388.e1 32981665 6 Tanner R.E. Brunker L.B. Agergaard J. Age-related differences in lean mass, protein synthesis and skeletal muscle markers of proteolysis after bed rest and exercise rehabilitation J Physiol 593 2015 4259 4273 26173027 7 Hvid L.G. Suetta C. Nielsen J.H. Aging impairs the recovery in mechanical muscle function following 4 days of disuse Exp Gerontol 52 2014 1 8 24447828 8 Burton E. Farrier K. Hill K.D. Codde J. Airey P. Hill A.M. Effectiveness of peers in delivering programs or motivating older people to increase their participation in physical activity: Systematic review and meta-analysis J Sports Sci 36 2018 666 678 28535358 9 Zhou X. Perez-Cueto F.J.A. Santos Q.D. A systematic review of behavioural interventions promoting healthy eating among older people Nutrients 10 2018 128 29373529 10 Orrow G. Kinmonth A.L. Sanderson S. Sutton S. Effectiveness of physical activity promotion based in primary care: Systematic review and meta-analysis of randomised controlled trials BMJ 344 2012 e1389 22451477 11 Lodge M. Wegrich K. The rationality paradox of nudge: Rational tools of government in a world of bounded rationality Law Policy 38 2016 250 267 12 Mertens S. Herberz M. Hahnel Ulf J.J. Brosch T. The effectiveness of nudging: A meta-analysis of choice architecture interventions across behavioral domains Proc Natl Acad Sci U S A 119 2022 e2107346118 13 Hallsworth M. Berry D. Sanders M. Stating Appointment Costs in SMS Reminders Reduces Missed Hospital Appointments: Findings from Two Randomised Controlled Trials PLoS One 10 2015 e0137306 26366885 14 Kahneman D. Tversky A. Prospect theory: An analysis of decision under risk Econometrica 47 1979 263 15 Levitt S.D. List J.A. Neckermann S. Sadoff S. The behavioralist goes to school: Leveraging behavioral economics to improve educational performance Am Econ J Econ Policy 8 2016 183 219 16 Laibson D. Golden Eggs and Hyperbolic Discounting 1997 Q J Econ 17 Milkman K.L. Beshears J. Choi J.J. Laibson D. Madrian B.C. Using implementation intentions prompts to enhance influenza vaccination rates Proc Natl Acad Sci U S A 108 2011 10415 10420 21670283 18 Arno A. Thomas S. The efficacy of nudge theory strategies in influencing adult dietary behaviour: A systematic review and meta-analysis BMC Public Health 16 2016 676 27475752 19 Sasaki S. Kurokawa H. Ohtake F. Effective but fragile? Responses to repeated nudge-based messages for preventing the spread of COVID-19 infection Jpn Econ Rev 72 2021 371 408 20 Yamada Y. Uchida T. Shiose T. Learning Health System in a Senior Retirement Community: A platform to promote implementation research Gerontol Geriatr Med 6 2020 1 6 21 Sewo Sampaio P.Y. Sampaio R.A. Yamada M. Arai H. Systematic review of the Kihon Checklist: Is it a reliable assessment of frailty? Geriatr Gerontol Int 16 2016 893 902 27444395 22 Yeo I.K. A new family of power transformations to improve normality or symmetry Biometrika 87 2000 954 959 23 Banerjee S. John P. Nudge plus: Incorporating reflection into behavioral public policy Behav Public Policy 2021 1 16 10.1017/bpp.2021.6 24 Koivisto J. Malik A. Gamification for older adults: A systematic literature review Gerontologist 61 2021 e360 e372 32530026 25 Yamada Y. Nanri H. Watanabe Y. Prevalence of frailty assessed by fried and kihon checklist indexes in a prospective cohort study: Design and demographics of the Kyoto-kameoka longitudinal study J Am Med Dir Assoc 18 2017 733.e7 733.e15 26 Marteau T.M. Ogilvie D. Roland M. Suhrcke M. Kelly M.P. Judging nudging: Can nudging improve population health? BMJ 342 2011 d228 21266441 27 Van Rookhuijzen M. De Vet E. Adriaanse M.A. The effects of nudges: One-shot only? Exploring the temporal spillover effects of a default nudge Front Psychol 12 2021 683262 34589018 28 Venkatesh V. Morris M.G. Why don’t men ever stop to Ask for Directions? Gender, social influence, and their role in technology acceptance and usage behavior Miss Q 24 2000 115 29 Waters L.A. Galichet B. Owen N. Eakin E. Who participates in physical activity intervention trials? J Phys Act Health 8 2011 85 103 21297189 30 Gilmour H. Social participation and the health and well-being of Canadian seniors Health Rep 23 2012 23 32

0

PMC9754120

NO-CC CODE

2022-12-16 23:26:27

no

J Am Med Dir Assoc. 2022 Dec 15; doi: 10.1016/j.jamda.2022.11.009

utf-8

J Am Med Dir Assoc

10.1016/j.jamda.2022.11.009

oa_other

==== Front J Affect Disord J Affect Disord Journal of Affective Disorders 0165-0327 1573-2517 Elsevier B.V. S0165-0327(22)00474-8 10.1016/j.jad.2022.04.130 Review Article Social anxiety and behavioral assessments of social cognition: A systematic review Alvi Talha Kumar Divya Tabak Benjamin A. ⁎ Department of Psychology, Southern Methodist University, 6116 N. Central Expressway, Suite 1300, Dallas, TX, USA ⁎ Corresponding author at: Department of Psychology, Southern Methodist University, 6116 N. Central Expressway, Suite 1300, Dallas, TX 75206, USA. 28 4 2022 15 8 2022 28 4 2022 311 1730 6 11 2021 17 4 2022 19 4 2022 © 2022 Elsevier B.V. All rights reserved. 2022 Elsevier B.V. Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. Background Social anxiety is highly prevalent and has increased in young adults during the COVID-19 pandemic. Since social anxiety negatively impacts interpersonal functioning, identifying aspects of social cognition that may be impaired can increase our understanding of the development and maintenance of social anxiety disorder. However, to date, studies examining associations between social anxiety and social cognition have resulted in mixed findings. Methods The aim of this systematic review was to summarize the literature on the association between social anxiety and social cognition, while also considering several potential moderators and covariates that may influence findings. Results A systematic search identified 52 studies. Results showed mixed evidence for the association between social anxiety and lower-level social cognitive processes (emotion recognition and affect sharing) and a trend for a negative association with higher-level social cognitive processes (theory of mind and empathic accuracy). Most studies examining valence-specific effects found a significant negative association for positive and neutral stimuli. Limitations Not all aspects of social cognition were included (e.g., attributional bias) and we focused on adults and not children, limiting the scope of the review. Conclusions Future studies would benefit from the inclusion of relevant moderators and covariates, multiple well-validated measures within the same domain of social cognition, and assessments of interpersonal functioning outside of the laboratory. Additional research examining the moderating role of attention or interpretation biases on social cognitive performance, and the potential benefit of social cognitive skills training for social anxiety, could inform and improve existing cognitive behavioral interventions. Keywords Social anxiety Social cognition Empathy Theory of mind Empathic accuracy Affect sharing ==== Body pmc1 Introduction Social anxiety disorder (SAD) is one of the most common mental illnesses (Kessler et al., 2005), and levels of social anxiety (SA) have increased in young adults since the onset of the COVID-19 pandemic (Hawes et al., 2021). SAD is characterized by a fear of evaluation and avoidance of social situations, which can negatively affect social functioning (American Psychiatric Association, 2013; Clark and Wells, 1995). Indeed, individuals with SAD report having more interpersonal problems and difficulty maintaining relationships (Davila and Beck, 2002; Kashdan et al., 2007; Tonge et al., 2020). These issues extend beyond close relationships and can have a substantial negative impact on occupational and educational functioning (Schneier et al., 1994; Wittchen et al., 2000) above and beyond the effects of comorbidities including depression (Aderka et al., 2012). One mechanism through which SA may negatively impact interpersonal functioning is via alterations in social cognitive processing (Morrison and Heimberg, 2013). Social cognitive ability (i.e., the ability to accurately understand others' thoughts and emotions; Brothers, 1990; Frith and Frith, 2007) has been associated with beneficial outcomes in interpersonal relationships beginning in childhood and extending into adulthood (Banerjee, 1997; Gleason et al., 2009; Sened et al., 2017). Given the negative impact of SA on interpersonal functioning (Alden and Taylor, 2004), a key predictor of health and well-being (Holt-Lunstad, 2021), identifying specific aspects of social cognition that may be impaired can increase our understanding of the development and maintenance of SAD. Although most studies using standardized measures of social cognition have included individuals with schizophrenia and autism spectrum disorder, an increasing amount of research has focused on anxiety symptomology, including SA. To date, three reviews and meta-analyses have examined the association between SA and social cognition in adults, but findings have been inconsistent. A potential reason for the lack of consistency stems from the complexity of social cognition and the extent to which different behavioral tasks index both related and distinct aspects of social cognitive processes. One method of categorizing the numerous domains of social cognition is to separate them into lower-level and higher-level processes (Green et al., 2013). Lower-level processes involve the recognition of basic social-emotional cues (e.g., emotion recognition; Green et al., 2008 and social cue perception; Penn et al., 2002; Pinkham et al., 2014), as well as affective empathic processes (i.e., shared feelings for others; Davis, 1983), such as affect sharing, which is thought to be automatic and reflexive (Hatfield et al., 1992; Singer and Lamm, 2009) and may precede cognitive processes (Fan and Han, 2008). Higher-level social cognitive processes involve the interpretation of complex social stimuli to make inferences about the thoughts and intentions of others (e.g., theory of mind; Baron-Cohen et al., 2001, and empathic accuracy; Ickes et al., 1990). In the first review and meta-analysis of the association between SA and social cognition, O’Toole et al. (2013) found a small negative effect between SA and basic emotion recognition and a moderate negative effect between SA and complex emotion recognition (higher-level social cognition). However, Plana et al. (2014) found no association between SA and emotion recognition, mentalizing ability, or social perception (no association with lower- or higher-level processes). More recently, Pittelkow et al. (2021) conducted a meta-analysis examining the association between SA and empathy (a multi-dimensional construct that is often used synonymously with the term social cognition). They distinguished cognitive empathy (i.e., the ability to recognize and identify others' emotions) from affective empathy (i.e., experiencing an emotion based on another's emotional experience), both of which can also be conceptualized as facets of social cognition. Pittelkow et al. (2021) found a positive association between SA and affective empathy and no association between SA and cognitive empathy. However, these results included studies measuring empathic processes based on self-report, which ask about one's perception of their own empathic tendencies or abilities, as well as behavioral assessments of social cognition, which examine the accuracy or congruence of one's responses in relation to targets or normed reference groups. The distinction between self-report and behavioral measures of social cognition is important given the weak correlation between the two, which suggests that self-report measures are not accurate proxies for ability (Murphy and Lilienfeld, 2019). Thus, it is important for reviews to distinguish between self-report and behavioral assessments when interpreting results to avoid conflating findings for separable constructs. Due in large part to a lack of studies available, previous reviews (O’Toole et al., 2013; Pittelkow et al., 2021; Plana et al., 2014) have also not examined the association between SA and empathic accuracy, a behavioral measure of higher-level social cognition. Beyond the type of social cognition assessed, previous reviews of SA and social cognition have not always examined valence-specific associations of SA on social cognitive processes or the method of measuring SA (i.e., dimensional measures, cutoff scores from dimensional measures, or diagnostic groups). Based on studies showing that greater levels of SA are associated with lower social cognitive ability for neutral and positive stimuli (Alvi et al., 2020; Washburn et al., 2016), parsing apart whether associations are valence-specific (between negative, neutral, and positive stimuli) may provide further clarity on associations between SA and social cognition that are not apparent when collapsing across all stimuli. Indeed, although Pittelkow et al. (2021) considered the potential moderating role of positive and negative valence, they did not examine the association between SA and social cognition for neutral stimuli. Finally, noting whether SA is assessed dimensionally or categorically may clarify previous findings. For example, when using the same facial recognition task, some have found no association between SA and emotion recognition when SA is assessed dimensionally (Mullins and Duke, 2004), while others have found a significant, positive association when SA is categorized based on cut-off scores (Hunter et al., 2009). Thus, considering the type of SA assessment may help to elucidate whether associations are specific to individuals with the highest levels of SA (i.e., those meeting criteria for SAD based on diagnostic interview), or if associations are evident when assessing SA dimensionally. In addition to moderators that may help explain mixed findings, the inclusion of statistical covariates is also important to consider. Depression (Bora and Berk, 2016), gender (Babchuk et al., 1985; Doherty et al., 1995; Magen and Konasewich, 2011; Thayer & Johnsen, 2000), age (Ruffman et al., 2008), alexithymia (Di Tella et al., 2020), and IQ (Brüne, 2003) have all been associated with social cognitive functioning and/or SA, and therefore represent potentially competing predictors that may account for associations between SA and social cognition. For example, previous work has shown no association between SA and certain forms of social cognition (i.e., affect sharing) when including age, gender, depressive symptoms, alexithymia, and target expressivity as covariates; however, the association becomes statistically significant when covariates are removed (Alvi et al., in press). As a result, prior reviews that have not compared findings from studies that accounted for relevant covariates with those that did not, prevents an understanding of how much these related factors may explain the associations shown between SA and social cognitive ability. For example, the most recent meta-analysis by Pittelkow et al. (2021) did not examine the extent to which studies accounted for covariates such as depression and alexithymia. 1.1 Review objectives The present systematic review assessed whether there is an association between SA and behavioral measures of social cognition, while also considering several potential moderators and covariates that may influence findings. The primary objectives were to determine trends in the literature based on the: 1) type of SA assessment, 2) type of social cognition assessment, 3) domain of social cognition, 4) valence of stimuli, and 5) inclusion of covariates. By providing a more fine-grained examination of the different domains of social cognition, and parsing results based on potential moderators and covariates, our goal was to help clarify the mixed associations between SA and social cognition that have been found in prior reviews. 2 Method The present study was designed in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement (Moher et al., 2015). 2.1 Literature search Articles were identified through searches in PsycInfo, Psych and Behavioral Sciences Collection, and PsycArticles from the earliest available dates through September 26th, 2020. In all databases, the following search string was used: (social anxiety OR social phobia) AND (emotion recognition OR emotion knowledge OR emotion processing OR theory of mind OR mentalizing OR mentalising OR social perception OR empathic accuracy OR cognitive empathy OR social cognition OR social cognitive ability OR affective empathy OR affect sharing). Searches were limited to peer-reviewed articles written in English. Forward and backward searching were used in selected articles to capture any additional relevant articles. 2.2 Inclusion and exclusion criteria Inclusion criteria included: a) an adult sample (≥18 years old), b) a measure of SA and/or a clinical, SA group, and c) at least one behavioral measure of social cognition. Based on meta-analytic evidence that self-report assessments of empathy are only weakly related to behavioral assessments of social cognition (Murphy and Lilienfeld, 2019), in the present review, we focus specifically on behavioral assessments. Exclusion criteria included: a) use of a child/adolescent sample, b) no measure of SA or SA group, c) a sample that included comorbid diagnoses, d) no behavioral measure of social cognition, and e) imaging studies with no behavioral outcome. We chose to exclude studies containing individuals with comorbid disorders who did not have a primary diagnosis of SA since the focus of our review is on the unique effects of SA on social cognitive abilities. Importantly, recent evidence has shown that many interpersonal difficulties experienced by individuals with SAD exist above and beyond the effects of comorbid depression symptoms (Tonge et al., 2020). 2.3 Task classification Social cognition tasks were classified into their respective domains using the framework proposed by Green et al. (2008). Emotion recognition tasks include those that prompt participants to identify a discrete emotion (e.g., happiness, sadness) in response to static or video stimuli (Heberlein et al., 2004; Joormann and Gotlib, 2006; Nowicki and Carton, 1993). Tasks measuring social perception include presentation of social scenes in which participants are asked to assess domains such as nonverbal cues, status, and intimacy (Conzelmann et al., 2013; Schroeder, 1995a, Schroeder, 1995b; Veljaca and Rapee, 1998). Measures of theory of mind include static and video stimuli that prompt participants to infer the mental states of others (Baron-Cohen, 1995; Baron-Cohen et al., 2001; Dziobek et al., 2006; Tibi-Elhanany et al., 2011). Studies in the present review assessed empathic accuracy for strangers using a video task in which participants continuously rate their perceptions of a target's emotional state as the target tells an autobiographical story (Kern et al., 2013). The targets also rate themselves, and the ratings of the perceiver and target are then correlated to create an empathic accuracy score. Studies in the present review assessed affect sharing using the same empathic accuracy video task with altered instructions that ask participants to rate their own feelings, as they watch the target individual (Morrison et al., 2016). All tasks are considered measures of cognitive empathy, apart from the affect sharing task, which measures affective empathy. 2.4 Screening and coding procedure Abstracts were independently screened by two raters using Abstrackr (http://abstrackr.cebm.brown.edu), an online software tool for screening and organizing literature searches. After the initial screen, full-texts were reviewed independently by both raters and assessed for inclusion criteria. Disagreements were discussed among raters and resolved through consensus. Studies were coded for the following: total sample, mean age, percent female, assessment of SA, assessment of social cognition task, statistical covariates included, and effect size. Studies were coded for all covariates included, and then subsequently coded for specific covariates of interest (depressive symptoms, gender, age, alexithymia, and IQ). For studies that reported effect sizes, coding included both the metric reported in the original paper and the equivalent using a common metric (Cohen's d) for ease of interpretation. Coding of all included full-text articles was completed by the first author. Further, 20% of the articles were randomly selected and coded by the second author to confirm agreement. Coders had 95% agreement prior to discussion and resolution of any disagreements. 3 Results 3.1 Search results and study selection An overview of study selection and flow can be seen in Fig. 1 . Database searches from PsycInfo, Psych and Behavioral Sciences Collection, and PsycArticles resulted in 1196 records. Following duplicate identification and removal (n = 148), 1048 unique records remained. An additional 16 records were identified through forward and backward searching, resulting in a total of 1064 records. After the first round of abstract screening, 935 records were excluded for the following reasons: no behavioral measure of social cognition (n = 436), no measure of SA or socially anxious group (n = 425), an imaging study with no behavioral outcomes (n = 34), child/adolescent samples (n = 33), case study (n = 7). The remaining 129 full-text articles were assessed for eligibility, and 81 articles were excluded for various reasons: no behavioral measure of social cognition (n = 67), no measure of SA or socially anxious groups (n = 7), samples with comorbid diagnoses (n = 4), child/adolescent samples (n = 2), an imaging study with no behavioral outcomes (n = 1). The final sample included 48 full-text articles.Fig. 1 PRISMA flow diagram Fig. 1 3.2 Study characteristics Across 48 full-text articles, 52 studies were identified. Since some of the articles included multiple studies, the number of social cognitive assessments and findings are sometimes greater than the total number of studies. Table 1 shows an overview of study characteristics. Most studies measured emotion recognition (n = 30), followed by theory of mind (n = 11), social perception (n = 5), empathic accuracy (n = 3), and affect sharing (n = 3). Sample size varied across studies (ranging from 20 to 1485 individuals), with an average total sample of 153.67 participants, and the direction and size of effects varied across studies as well (Table 2 ).Table 1 Overview of studies and study characteristics. Table 1 Total N (SA/HC) Mean age SA/HC % women SA/HC Sample type Statistical covariates Assessment of social anxiety Social cognition task Emotion recognition Alvi et al. (2020) - Study 1 1485 25.8 69.0 Student and Community Age, gender, depressive symptoms, neuroticism, extraversion, mentalizing, social anhedonia Composite (LSAS, SIAS, SPS) Emotion Perceptions of Biological Motion Task (Heberlein et al., 2004) Arrais et al. (2010) 231 (78/153) 22.3/21.4 61.5/65.4 Student None Diagnostic groups (based on SCID for DSM-IV) Facial emotion recognition task using stimuli from Pictures of Facial Affect (Ekman and Friesen, 1976) Auyeung and Alden (2020) -Study 1 134 20.5 79.8 Student Depression, age SIAS-S Video empathy task (with social exclusion manipulation) Auyeung and Alden (2020) -Study 2 126 (63/63) 29.3/29.9 60.3/61.9 Community Depression, age Diagnostic groups (based on ADIS for DSM-IV) Video empathy task (with social exclusion manipulation) Bell et al. (2011) 57 (30/27) 37/39 36.7/40.7 Community Gender, medication status Diagnostic groups (based on SCID for DSM-IV) Morphed faces task using stimuli from Pictures of Facial Affect (Ekman and Friesen, 1976) Bodner et al. (2012) 80 (39/41) 28.6 40.0 Community Depression, age Groups based on LSAS Vocal improvisation recognition task (Bodner and Gilboa, 2006) and vocal prosody recognition task (Berger, 2002) Button et al. (2013) 102 (52/50) 22/23 100/100 Community None Groups based on BFNE Facial emotion recognition task using stimuli from the Karolinska Directed Emotional Faces set (Lundqvist et al., 1998) Campbell et al. (2009) 40 (12/18) 31.9/30.4 58.3/35.7 Community None Diagnostic groups (based on MINI for DSM-IV) Facial emotion recognition task using stimuli from the Japanese and Caucasian Facial Expressions of Emotion set (Matsumoto and Ekman, 1988) Dickter et al. (2018) - Study 2 208 (51/63) 18.9 59.3 Student None Groups based on SPAI Facial emotion visual search task using stimuli from the NimStim Set of Facial Expressions (Tottenham et al., 2009) Garner et al. (2009) 33 (16/17) 43.1/39.9 68.8/47.1 Community None Diagnostic groups (based on MINI for DSM-IV) Morphed faces task using stimuli from the NimStim Set of Facial Expressions (Tottenham et al., 2009) Gilboa-Schechtman et al. (2008) - Study 2 202 (54/65) 29.9/27.6 51.9/52.3 Community None Diagnostic groups (based on SCID for DSM-IV) Morphed faces task using stimuli from database of facial expressions (Halberstadt and Niedenthal, 1997) Hagemann et al. (2016) 42 (21/21) 26.9/27.0 76.2/76.2 Community None Diagnostic groups (based on SCID for DSM-IV) Facial emotion recognition task using stimuli from FACES Database (Ebner et al., 2010), Karolinska Directed Emotional Faces (Lundqvist et al., 1998), Radboud Faces Database (Langner et al., 2010) and the NimStim Set of Facial Expressions (Tottenham et al., 2009) Heuer et al. (2010) 57 (27/30) 20.0/20.0 100/100 Student None Groups based on LSAS Morphed faces task using stimuli from the Karolinska Directed Emotional Faces set (Lundqvist et al., 1998) Hunter et al. (2009) 158 (24/121) 18.71 52.7 Student Gender, depression, trait anxiety Groups based on SPS Diagnostic Analysis of Non-verbal Accuracy (Nowicki and Carton, 1993; Nowicki et al., 1998) M. Jacobs et al. (2008) 49 (28/21) 32.4/36.0 46.4/47.6 Community General anxiety, overall impairment Diagnostic groups (based on SCID for DSM-IV) Mayer-Salovey-Caruso Emotional Intelligence Test- Experiential Emotional Intelligence (Mayer et al., 2002) Joormann and Gotlib (2006) 72 (26/25) 30.2/31.6 61.5/68 Community Error rate Diagnostic groups (based on SCID for DSM-IV) Morphed faces task using stimuli from the Facial Expressions of Emotions-Stimuli and Test set (Young et al., 2002) Lau et al. (2014) 264 College-aged 69.3 Student Gender, neuroticism SAS-A Video Emotion Recognition Task (Kang and Lau, 2013) Mohlman et al. (2007) 52 (26/26) 21.5/21.1 62/65 Student Depression, trait anxiety, stimulus order Diagnostic groups (based on SCID for DSM-IV) Facial emotion recognition task (with anxiety manipulation) using stimuli from Horstmann (2002) Montagne et al. (2006) 50 (24/26) 36.7/37.6 58.3/53.8 Community None Diagnostic groups (based on MINI for DSM-IV) Emotion Recognition Task (Montagne et al., 2007) Mullins and Duke (2004) 70 19.2 100 Student Pre-test anxiety FNES Diagnostic Analysis of Non-verbal Accuracy (Nowicki and Carton, 1993) (with anxiety manipulations) Oh et al. (2018) 112 (56/56) 27.3/25.8 46.4/44.6 Community None Diagnostic groups (based on MINI for DSM-IV) Facial emotion recognition task using stimuli from Pictures of Facial Affect (Ekman and Friesen, 1976) Phan et al. (2006) 20 (10/10) 26.7/26.6 50.0/50.0 Unknown None Diagnostic groups (based on SCID for DSM-IV) Facial emotion recognition task using Penn Emotion Recognition set (Gur et al., 2002) Philippot and Douilliez (2005) 80 (21/39) 30.0/33.0 33.3/51.3 Community None Diagnostic groups (based on MINI for DSM-IV) Morphed faces task using stimuli from Hess and Blairy (1995) Quadflieg et al. (2007) 30 (15/15) 23.3/23.9 53.3/53.3 Student Depressive symptoms Diagnostic groups (based on SCID for DSM-IV) Emotion prosody recognition task using stimuli from the “Magdeburger Prosodie-Korpus” (Wendt and Scheich, 2002) Schofield et al. (2007) 100 (49/51) 18.5/18.9 75.5/52.9 Student None Groups based on BFNE Morphed faces task using stimuli from the Japanese and Caucasian Facial Expressions of Emotion set (Matsumoto and Ekman, 1988) Straube et al. (2004) 20 (10/10) 25.0/23.2 60.0/60.0 Student and Community None Diagnostic groups (based on SCID for DSM-IV) Facial emotion recognition task using stimuli from the NimStim Set of Facial Expressions (Tottenham et al., 2009 Toro-Alves et al. (2016) 43 (22/21) 20.5/22.5 40.9/42.9 Student Age Groups based on SPIN Morphed faces task using stimuli from the NimStim Set of Facial Expressions (Tottenham et al., 2009 Tseng et al. (2017) 62 (31/31) 29.6/30.9 45.2/45.2 Community Depression Diagnostic groups (based on MINI for DSM-IV) Diagnostic Analysis of Non-verbal Accuracy 2- Taiwan version (Chen, 2006) Winton et al. (1995) 24 (13/11) 20.6/22.7 69.2/54.5 Student None Groups based on FNES Facial emotion recognition using stimuli from the Japanese and Caucasian Facial Expressions of Emotion set (Matsumoto and Ekman, 1988) and the Nonverbal Discrepancy Test (DePaulo, 1978) Yoon et al. (2007) 22 (11/11) 27.0/26.9 54.5/54.5 Unknown None Diagnostic groups (based on SCID for DSM-IV) Facial emotion recognition task using Penn Emotion Recognition set (Gur et al., 2002) Social perception Hampel et al. (2011) 110 31.5 60.0 Community Trait anxiety SBQ, SCQ, SAQ, and SAM Magdeburg Test of Social Intelligence- Social Perception (Conzelmann et al., 2013) Schroeder (1995a) 84 22.0 59.5 Student None SCS Interpersonal Perception Task (Costanzo and Archer, 1989) Schroeder (1995b) 68 18.8 52.9 Student None SCS Interpersonal Perception Task (Costanzo and Archer, 1989) Schroeder and Ketrow (1997) 161 18–31 55.3 Student None PRCA Interpersonal Perception Task (Costanzo and Archer, 1989) Veljaca and Rapee (1998) 39 (19/20) 22.4/24.2 89.5/70.0 Student None Groups based on APPQ and NOQ Social cue detection task Affect sharing Alvi et al. (in press) - Study 1 202 19.8 68.2 Student Age, gender, depressive symptoms, alexithymia, target expressivity Composite (LSAS, SIAS, SPS) Affect sharing video task adapted from Kern et al. (2013) Alvi et al. (in press) - Study 2 324 37.1 45.4 Community Age, gender, depressive symptoms, alexithymia, target expressivity Composite (LSAS, SIAS, SPS) Affect sharing video task adapted from Kern et al. (2013) Morrison et al. (2016) 64 (32/32) 31.9/31.7 43.7/43.7 Community None Diagnostic groups (based on ADIS for DSM-IV) EC video task adapted from Zaki et al. (2008) Theory of mind Alvi et al. (2020) - Study 1 1485 25.8 69.0 Student and Community Age, gender, depressive symptoms, neuroticism, extraversion, mentalizing, social anhedonia Composite (LSAS, SIAS, SPS) Reading the Mind in the Eyes Test (Baron-Cohen et al., 2001) Ballespí et al. (2018) 113 (33/80) 21.1 85.8 Student None Groups based on SPAI-B and BFNE Movie for the Assessment of Social Cognition task (Dziobek et al., 2006) and induced mentalization in self-referential paradigm Buhlmann et al. (2015) 140 (35/35) 32.2/32.7 60.0/60.0 Community Non-social inferencing (control items) Diagnostic groups (based on SCID for DSM-IV) Movie for the Assessment of Social Cognition task (Dziobek et al., 2006) Hezel and McNally (2014) 80 (40/40) 26.5/20.1 67.5/85 Student and Community Gender Diagnostic groups (based on MINI for DSM-IV) Reading the Mind in the Eyes Test (Baron-Cohen et al., 2001) and Movie for the Assessment of Social Cognition task (Dziobek et al., 2006) M. Jacobs et al. (2008) 49 (28/21) 32.4/36.0 46.4/47.6 Community General anxiety, overall impairment Diagnostic groups (based on SCID for DSM-IV) Mayer-Salovey-Caruso Emotional Intelligence Test- Strategic Emotional Intelligence (Mayer et al., 2002) Lenton-Brym et al. (2018) 113 (78/35) 19.4/20.4 71.8/54.3 Student None Groups based on SPIN Reading the Mind in the Eyes Test (Baron-Cohen et al., 2001) and Movie for the Assessment of Social Cognition task (Dziobek et al., 2006) Lyvers et al. (2019) 242 23.2 61.6 Student and Community Age, gender, depressive symptoms, alexithymia SIAS Reading the Mind in the Eyes Test (Baron-Cohen et al., 2001) Maleki et al. (2020) 107 (35/35) 27.5/28.4 54.3/48.6 Community None Diagnostic groups (based on SCID for DSM-IV) Reading the Mind in the Eyes Test (Baron-Cohen et al., 2001) and faux pas task (Baron-Cohen et al., 1999) Sutterby et al. (2012) 56 (27/29) 19.1 59.3/62.1 Student None Groups based on SPAI Reading the Mind in the Eyes Test (Baron-Cohen et al., 2001) and the Awareness of Social Inferences Test (McDonald et al., 2006) Tibi-Elhanany and Shamay-Tsoory (2011) 43 (45/22) 25.3/25 71.4/45.5 Community Control condition Groups based on LSAS Cartoon ToM task (Baron-Cohen, 1995) Washburn et al. (2016) 119 (12/43) 19.8/18.7 58.3/65.0 Student and Community Age, gender, ethnicity, education, response latency Diagnostic groups (based on SCID for DSM-IV) Reading the Mind in the Eyes Test (Baron-Cohen et al., 2001) and Movie for the Assessment of Social Cognition task (Dziobek et al., 2006) Empathic accuracy Alvi et al. (2020) - Study 2 390 19.6 77.2 Student Gender, depressive symptoms, neuroticism, extraversion, mentalizing, social anhedonia, target expressivity, video order Composite (LSAS, SIAS, SPS) EA video task (Kern et al., 2013) Auyeung and Alden (2016) 121 20.1 78.7 Student Age SIAS EA video task (with anxiety induction) Morrison et al. (2016) 64 (32/32) 31.9/31.7 43.7/43.7 Community None Diagnostic groups (based on ADIS for DSM-IV) EA video task (Zaki et al., 2008) Note. LSAS = Liebowitz Social Anxiety Scale, SIAS = Social Interaction Anxiety Scale, SIAS-S = Social Interaction Anxiety Scale-Straight-forward Score, SPS = Social Phobia Scale, SCID = Structured Clinical Interview for DSM-IV, ADIS = Anxiety Disorders Interview Schedule for DSM-IV, BFNE = Brief Fear of Negative Evaluation scale, MINI = Mini International Neuropsychiatric Interview for DSM-IV, SPAI = Social Phobia and Anxiety Inventory, SAS-A = Social Anxiety Scale for Adolescents, FNES = Fear of Negative Evaluation Scale, SPIN = Social Phobia Inventory, SBQ = Social Behavior Questionnaire, SCQ = Social Cognitions Questionnaire, SAQ = Social Attitudes Questionnaire, SCS = Self-consciousness Scale, PRCA = Personal Report of Communication Apprehension, APPQ = Albany Panic and Phobia Questionnaire, NOQ = Negative Outcome Questionnaire, ToM = theory of mind, EA = empathic accuracy. Table 2 Study findings and effect sizes. Table 2 Effect size Finding Emotion recognition Alvi et al. (2020) - Study 1 For total, d = 0.47; for negative, d = 0.44; for positive, d = 0.36; for neutral d = 0.40 No association between SA and emotion recognition Arrais et al. (2010) Not reported Group X gender interaction- women in SA group required less emotional intensity to recognize faces displaying fear, happiness, and sadness compared to HC Auyeung and Alden (2020) - Study 1 Not reported Across both conditions, higher SA predicted greater accuracy of target’s negative affect Auyeung and Alden (2020) - Study 2 d = 0.55 Across both conditions, SAD group had greater accuracy of target negative affect compared to HC Bell et al. (2011) Not reported No differences between groups on accuracy Bodner et al. (2012) d = 0.94 SA group was less accurate at recognition of happy voices in female voices compared to HC Button et al. (2013) Not reported No difference between groups on accuracy Campbell et al. (2009) Not reported No difference between groups on accuracy Dickter et al. (2018) - Study 2 d = 0.37 For complex faces, high SA group was less accurate than low SA group Garner et al. (2009) d = 0.64 SA group was less accurate at identifying fearful faces than HC Gilboa-Schechtman et al. (2008) - Study 2 d = 0.55 SA group less sensitive to angry expressions than HC Hagemann et al. (2016) d = 0.55 No difference between groups on accuracy Heuer et al. (2010) d = 0 No difference between groups on accuracy Hunter et al. (2009) Not reported SA group was more accurate for happy, sad, and fearful faces compared to HC M. Jacobs et al. (2008) d = 1.62 No difference between groups on accuracy; negative correlation between SA and experiential EI Joormann and Gotlib (2006) Not reported No difference between groups on accuracy; SA group was able to identify angry faces more quickly than HC Lau et al. (2014) d = 0.84 Lower emotion recognition associated with lower social anxiety Mohlman et al. (2007) d = 1.66 (for positive association); d = 1.96 (for negative association) SA group was more accurate at identifying angry faces, and less accurate at neutral faces, than HC, following threat manipulation Montagne et al. (2006) Not reported SA group was less sensitive to negative emotions compared to HC Mullins and Duke (2004) Not reported No association between SA and accuracy, regardless of condition Oh et al. (2018) Not reported SA group was less accurate for total, fear, surprise, neutral, and happy stimuli than HC Phan et al. (2006) Not reported No difference between groups on accuracy Philippot and Douilliez (2005) d = 0.34 No difference between groups on accuracy Quadflieg et al. (2007) Not reported SA group was less accurate for happy utterances, but more accurate for fearful and sad utterances, than HC Schofield et al. (2007) Not reported No differences between groups on accuracy Straube et al. (2004) Not reported No differences between groups on accuracy Torro-Alves et al. (2016) Not reported SA group was more accurate for angry faces with 25% intensity than HC Tseng et al. (2017) d = 0.52 (across stimuli) and d = 0.67 (for fear accuracy) HC had greater accuracy than SA group across facial and vocal accuracy; SA group was less accurate in fear accuracy for faces Winton et al. (1995) Not reported SA group was more accurate at identifying negative facial expressions, and less accurate at neutral expressions, than HC; no difference in groups for videos (facial/body and auditory stimuli) Yoon et al. (2007) Not reported No difference between groups on accuracy Social perception Hampel et al. (2011) d = 0.52 Negative association between SA and social perception Schroeder (1995a) d = 0.47 No association between SA and interpersonal perception Schroeder (1995b) d = 0.49 Negative association between SA and interpersonal perception Schroeder and Ketrow (1997) d = 0.54 Negative association between SA and interpersonal perception Veljaca and Rapee (1998) Not reported SA group was more accurate in negative social cue detection, and less accurate at positive cue detection, compared to HC Affect sharing Alvi et al. (in press) - Study 1 d = 0.04 No effect of SA on affect sharing Alvi et al. (in press) - Study 2 d = 0.02 No effect of SA on affect sharing Morrison et al. (2016) d = 0.55 SA group had less congruence for positive stimuli, compared to HC Theory of mind Alvi et al. (2020) - Study 1 For total, d = 0.86; for negative, d = 0.77; for positive, d = 0.44; for neutral d = 0.71 Negative association between SA and accuracy for total, negative, positive, and neutral stimuli Ballespí et al. (2018) Not reported No difference between groups on accuracy in video task; hyper-mentalizing in SA group for self-referential paradigm Buhlmann et al. (2015) d = 0.72 SA group was less accurate than HC in ToM Hezel and McNally (2014) d = 0.61 (for videos task) and d = 0.70 (for eyes task) SA group was less accurate on both the video and eyes tasks compared to HC M. Jacobs et al., (2008) d = 0.12 No difference between groups on accuracy Lenton-Brym et al. (2018) d = 0.06 No difference between groups on accuracy Lyvers et al. (2019) d = 0.45 Negative association between SA and ToM Maleki et al. (2020) Not reported SA group had lower ToM on eyes task compared to HC; for positive and neutral valence, HC had greater ToM than SAD group; for negative valence, HC had lower ToM than SAD group; No difference between groups on faux pas accuracy Sutterby et al. (2012) d = 0.67 Group X gender interaction for tasks; women in SA group were more accurate compared to women in HC group Tibi-Elhanany et al. (2011) Not reported SA group had greater cognitive accuracy, but less affective accuracy, than HC Washburn et al. (2016) d = 0.59 SA group had lower ToM on eyes task for total, positive, and neutral stimuli compared to HC; no effect for video task Empathic accuracy Alvi et al. (2020)- Study 2 Not reported Higher negative association between SA and positive, compared to negative, stimuli Auyeung and Alden (2016) Not reported Group X condition interaction; SA associated greater accuracy for target negative affect when faced with social threat Morrison et al. (2016) d = 0 No difference between groups on accuracy Note. Effect sizes were converted to Cohen's d to compare effect sizes using a standard unit; SA = social anxiety; SAD = social anxiety disorder; HC = healthy controls; ToM = theory of mind. 3.3 Lower-level social cognition 3.3.1 Emotion recognition There were no apparent trends in findings from studies examining the association between SA and emotion recognition. Results based on SA assessment or emotion recognition assessment were also mixed apart from video emotion recognition and vocal/prosody tasks. Valence-specific findings were also mixed, however, there was some consistency with statistically significant associations of SA on neutral stimuli (i.e., all studies found greater SA to be associated with decreased accuracy for neutral stimuli). Of the 30 studies examining the association between SA and emotion recognition, 16 studies found a statistically significant association and 14 studies found a non-significant association. Of the studies that noted statistical significance, eight found a positive association (i.e., higher levels of SA were related to greater emotion recognition ability) and 11 studies found a negative association (higher levels of SA were related to lower emotion recognition ability). In studies that reported an effect size, values ranged from d = 0–1.96, with an average effect size of d = 0.72. When examining statistically significant valence-specific associations, for negative stimuli, six studies found a positive association, with higher levels of SA associated with greater accuracy in identifying stimuli as negative (Auyeung and Alden, 2020; Hunter et al., 2009; Mohlman et al., 2007; Quadflieg et al., 2007; Torro-Alves et al., 2016; Winton et al., 1995), and four studies found a negative association, where higher levels of SA were related to less accuracy (Garner et al., 2009; Gilboa-Schechtman et al., 2008; Oh et al., 2018; Tseng et al., 2017). For positive stimuli, one study found a positive association (Hunter et al., 2009) and three studies found a negative association (Bodner et al., 2012; Oh et al., 2018; Quadflieg et al., 2007). Studies that examined neutral stimuli had more consistency in findings; all three found a negative association between SA and emotion recognition for neutral stimuli (Mohlman et al., 2007; Oh et al., 2018; Winton et al., 1995). As mentioned previously, most studies used diagnostic groups of socially anxious individuals and healthy controls. In these studies, nine found a statistically significant association and nine found a non-significant association. Studies that created groups based on dimensional assessment also had mixed results, with five finding a statistically significant association (Bodner et al., 2012; Dickter et al., 2018; Hunter et al., 2009; Torro-Alves et al., 2016; Winton et al., 1995) and three finding a non-significant association (Button et al., 2013; Heuer et al., 2010; Schofield et al., 2007). Finally, in studies using dimensional assessment of SA, two found a statistically significant association (Auyeung and Alden, 2020; Lau et al., 2014) and two found a non-significant association (Alvi et al., 2020; Mullins and Duke, 2004). Thus, there was no clear trend in findings based on category of SA assessment. In considering the type of emotion recognition assessment, in studies that used a static, facial emotion recognition task, seven studies found a statistically significant association and eight studies found a non-significant association. Similarly, in studies that utilized a morphed faces task, half reported a statistically significant association (Garner et al., 2009; Gilboa-Schechtman et al., 2008; Montagne et al., 2006; Torro-Alves et al., 2016) and half reported a non-significant association (Heuer et al., 2010; Joormann and Gotlib, 2006; Philippot and Douilliez, 2005; Schofield et al., 2007). There was more consistency in results for studies that used a video emotion recognition task, as three studies found a statistically significant association (Auyeung and Alden, 2020; Lau et al., 2014) and one found a non-significant association (Winton et al., 1995). In addition, all three studies using a vocal/prosody task found a statistically significant association (Bodner et al., 2012; Quadflieg et al., 2007; Tseng et al., 2017). There was only one study that used an emotional perception of biological motion task (Alvi et al., 2020) and one study that used an emotion intelligence task (M. Jacobs et al., 2008); neither found a statistically significant association of SA on emotion recognition. 3.3.2 Social perception Overall, most studies found a statistically significant negative association between SA and social perception (i.e., greater SA was associated with decreased social perception). Although findings with social perception were generally consistent, there were few studies that examined this domain of social cognition. In studies that found a statistically significant association, four studies found a negative association, with higher levels of SA relating to reduced social perception (Hampel et al., 2011; Schroeder, 1995b; Schroeder and Ketrow, 1997; Veljaca and Rapee, 1998) and one study found a positive association between higher levels of SA and greater social perception (Veljaca and Rapee, 1998). Only one study found a non-significant association between SA and social perception (Schroeder, 1995a). For studies that reported an effect size, values ranged from d = 0.47–0.75, with an average effect size of d = 0.51. Only one study examined valence with a moderator, finding a significant negative association between SA and social perception for positive stimuli, such that those with higher levels of SA had lower accuracy in social perception (Veljaca and Rapee, 1998). Of the studies that used a dimensional measure of SA, the majority found a statistically significant association (Hampel et al., 2011; Schroeder, 1995b; Schroeder and Ketrow, 1997) and one study did not (Schroeder, 1995a). Further, the study that created groups based on dimensional assessment of SA also found a statistically significant association (Veljaca and Rapee, 1998). As most studies found a significant association, findings by measures of social perception were consistent. The studies that utilized a social intelligence task and social cue detection task found significant results. The studies that used the interpersonal perception task had mixed findings; two studies found a statistically significant association (Schroeder, 1995b; Schroeder and Ketrow, 1997) and one did not (Schroeder, 1995a). Thus, overall, most studies examining SA and social perception found a significant, negative association. 3.3.3 Affect sharing There were no apparent trends in findings in studies examining the association between SA and behaviorally assessed affect sharing based on statistically significant associations, type of SA assessment, or valence of stimuli. In the three studies that examined the association between SA and affect sharing, two studies found a non-significant association (Alvi et al., in press) and one study found a statistically significant negative association (higher levels of SA were related to lower affect sharing) that was moderated by valence (Morrison et al., 2016). Specifically, SA was negatively associated with affect sharing for positive stimuli (Morrison et al., 2016). Effect sizes ranged from d = 0.02–0.55, with an average effect size of d = 0.20. 3.4 Higher-level social cognition 3.4.1 Theory of mind Most studies examining the association between SA and theory of mind found a statistically significant, negative association (i.e., higher levels of SA were associated with decreased theory of mind). Valence-specific findings were also consistent regarding positive and neutral stimuli (i.e., higher levels of SA were associated with reduced theory of mind for stimuli with positive and neutral valence). Results based on social cognitive assessment were also consistent in that nearly all studies that assessed theory of mind based on the eye region found a statistically significant association. Overall, most studies found a statistically significant association (n = 8), with six studies finding a negative association indicating that higher levels of SA were related to reduced theory of mind ability (Alvi et al., 2020; Buhlmann et al., 2015; Hezel and McNally, 2014; Lyvers et al., 2019; Maleki et al., 2020; Washburn et al., 2016), and three studies finding a positive association, with higher levels of SA relating to better theory of mind ability (Sutterby et al., 2012; Maleki et al., 2020; Tibi-Elhanany et al., 2011). Further, four studies found a non-significant association between SA and theory of mind (Ballespí et al., 2018; M. Jacobs et al., 2008; Lenton-Brym et al., 2018; Maleki et al., 2020). For studies that reported an effect size, values ranged from d = 0.06–0.86, with an average effect size of d = 0.53. Findings by valence were mixed, although most studies found a statistically significant negative association (greater SA was related to reduced theory of mind ability). For negative stimuli, two studies found a statistically significant negative association (Alvi et al., 2020; Washburn et al., 2016) and one study found a statistically significant positive association between higher levels of SA and greater theory of mind ability (Maleki et al., 2020). Three studies, which examined both positive and neutral stimuli, found that higher levels of SA levels were associated with reduced theory of mind (Alvi et al., 2020; Maleki et al., 2020; Washburn et al., 2016). Of the studies that used diagnostic groups, four found a statistically significant association (Buhlmann et al., 2015; Hezel and McNally, 2014; Maleki et al., 2020; Washburn et al., 2016) and one found a non-significant association (M. Jacobs et al., 2008). Studies that created groups based on dimensional assessment had mixed results, with two studies finding a statistically significant association (Ballespí et al., 2018; Tibi-Elhanany et al., 2011) and two studies finding a non-significant association (Lenton-Brym et al., 2018; Sutterby et al., 2012). The two studies that used dimensional assessment of SA were consistent, with both finding a statistically significant association (Alvi et al., 2020; Lyvers et al., 2019). The majority of studies used a theory of mind task based on eye-region, and nearly all found a statistically significant association (Alvi et al., 2020; Hezel and McNally, 2014; Lyvers et al., 2019; Maleki et al., 2020; Sutterby et al., 2012; Washburn et al., 2016), except one (Lenton-Brym et al., 2018). Studies that used video-based theory of mind tasks had mixed results, as three studies found a statistically significant association (Buhlmann et al., 2015; Hezel and McNally, 2014; Sutterby et al., 2012) and three studies did not (Ballespí et al., 2018; Lenton-Brym et al., 2018; Washburn et al., 2016). There was only one study that used a strategic emotional intelligence task (M. Jacobs et al., 2008) and one study that used a faux pas task (Maleki et al., 2020); neither found a statistically significant association of SA. Finally, one study used a cartoon theory of mind task and found a significant association between SA and theory of mind (Tibi-Elhanany et al., 2011). 3.4.2 Empathic accuracy There were only a few studies examining empathic accuracy, and findings were mixed. There were no apparent trends in findings based on statistically significant associations, type of SA assessment, or valence of stimuli. Of the three studies that assessed empathic accuracy, two studies found a statistically significant association between SA and empathic accuracy (Alvi et al., 2020; Auyeung and Alden, 2016) and one study found a non-significant association (Morrison et al., 2016). However, the study that found a non-significant association between SA and empathic accuracy had a smaller sample (n = 64) compared to the other two (ns = 121–390); thus, these results should be interpreted with caution. Of the studies that found a statistically significant association, one study found that higher levels of SA were related to reduced empathic accuracy (Alvi et al., 2020) and the other found the opposite: higher levels of SA were related to increased empathic accuracy (Auyeung and Alden, 2016). Findings by valence were also mixed. One study found that higher levels of SA were related to reduced empathic accuracy for positive stimuli (Alvi et al., 2020). In contrast, another study found that higher levels of SA were associated with greater empathic accuracy for negative stimuli (Auyeung and Alden, 2016). Further, the study that utilized diagnostic groups found a non-significant association (Morrison et al., 2016) whereas the two studies that assessed SA dimensionally did find a statistically significant association (Alvi et al., 2020; Auyeung and Alden, 2016). 3.5 Covariates Twenty-six (50.00%) studies included at least one statistical covariate in their model examining the association between SA and social cognition. Of the studies that did include covariates, 17 (65.38%) included two or less covariates. Across these studies, statistical covariates included age (11 studies; 42.31%), gender (11 studies; 42.31%), depression (13 studies; 50.00%), alexithymia (3 studies; 11.54%), and no studies statistically controlled for IQ. Most studies included age and gender as covariates, and several included depression; yet for all three variables, there was no clear trend in the patterns of results that included nonsignificant and significant associations in both the positive and negative directions (i.e., with higher levels of SA relating to both increased and decreased social cognitive ability) and effect sizes ranging from small to large. In the only study that controlled for alexithymia when examining SA and theory of mind, a negative association was found with a moderate effect size wherein individuals with higher levels of SA had reduced theory of mind ability (Lyvers et al., 2019). Interestingly, among the three studies examining SA and affect sharing, a significant association was only found for the one study that did not control for alexithymia or depression (Morrison et al., 2016), whereas the other two studies that did control for these variables found no association (Alvi et al., in press). These results suggest that covariates, particularly depression and alexithymia, may be most relevant when studying affect sharing; however, future studies are needed to replicate these findings. 3.6 Trends in results Most studies used diagnostic groups based on clinical interviews for the DSM-IV (45.9%) or cut-off scores through dimensional measures (27.9%) in their assessment of SA. Only 26.2% of studies used dimensional assessment of SA in their analyses. Overall, there was no general trend associated with the type of SA assessment and findings were mixed. Most studies examining social perception (80%), theory of mind (66.6%), and empathic accuracy (66.6%) found a statistically significant association. However, these domains, especially social perception and empathic accuracy, had far fewer studies than others, such as emotion recognition. Findings related to lower-level processes (emotion recognition, social perception, and affect sharing) were mixed. When these domains were collapsed, no general trend was found, which was largely driven by the heterogeneity in findings with emotion recognition. For studies examining higher-level processes (theory of mind and empathic accuracy), the majority (66.6%) found a statistically significant association. Further, most of these studies (63.7%) found a negative association between SA and higher-level social cognition, with those with higher levels of SA exhibiting reduced abilities. Among studies that found a valence-specific effect, most found negative associations for positive (90%) and neutral (100%) stimuli, whereas findings for negative stimuli were mixed. In the two studies that examined gender moderation, both found a positive association between SA and social cognition (emotion recognition and theory of mind) for females only. Among the four studies that included experimental anxiety inductions (Table 1), results were mixed and no trend in findings could be identified based on the type of study design. 4 Discussion This systematic literature review synthesized studies in adults that investigated the association between SA and behaviorally assessed social cognition. Further, it considered trends across studies based on the type of SA and social cognition assessment, the domain of social cognition, relevant covariates, and the valence of stimuli. The main finding of this review was that most studies examining the association between SA and higher-level social cognition found a statistically significant effect, with the majority finding a negative association (higher levels of SA relating to less social cognitive ability). These findings are consistent with O’Toole et al. (2013) who found a negative association between SA and complex emotions (higher-level social cognition) and suggest that greater SA is associated with impairment in higher-level social cognition. Findings related to lower-level social cognition were mixed, although the majority of studies examining the domain of social perception found a statistically significant negative effect of SA. Most studies did not examine valence-specific effects, but among those that did, findings suggest SA is negatively associated with the ability to accurately perceive positive and neutral stimuli. The mixed findings associated with lower-level social cognitive processes may be explained by several factors. First, the assessments used to measure lower-level processes (e.g., emotion recognition) have been more varied than the assessments used to measure other domains of social cognition; this increased variability may explain the differences in findings. Indeed, the majority of studies assessing higher-level processes used the same measures of social cognition (81% of studies examining theory of mind utilized one or both of the same tasks, and all of the studies examining empathic accuracy utilized a version of the same task). In contrast, emotion recognition tasks varied greatly in terms of the stimuli presented (i.e., the assessment of facial, vocal, or body cues), as well as the sensory modality (i.e., videos, pictures, sounds). Further, the greater number of studies that examined emotion recognition relative to other forms of social cognition, in conjunction with the different forms of assessment, contributes to this variability. The discrepancy in the number of studies between domains of social cognition may be due in part to feasibility. Most studies examining emotion recognition used a static emotion recognition task, which are easily administered, scored, and analyzed. In comparison, behavioral measures of empathic accuracy and affect sharing often require specific computer software to run and analyze data, greater time for administration, and advanced statistical techniques (e.g., multilevel modeling). Finally, the mixed findings with lower-level social cognition may also be explained by the infrequent inclusion of covariates. The majority of studies examining these domains either included no covariates, or only a few (only 17.3% of all studies included more than two covariates), and these were generally limited to age, gender, and depressive symptoms. Future studies are needed to systematically examine the extent to which the factors we have identified contribute to increased replicability (e.g., do findings replicate using the same sample and covariates but with different measures of emotion recognition). Evidence for a negative association between SA and positive or neutral (but not negative) stimuli may be explained by socially anxious individuals' attentional biases away from positive information (Taylor et al., 2010), and the tendency to suppress positive emotions (Farmer and Kashdan, 2012), which can result in the misinterpretation of these social cues. Further, socially anxious individuals tend to interpret ambiguous stimuli more negatively (Morrison and Heimberg, 2013), which may explain the negative association between SA and accuracy for neutral cues. Future studies should include measures of cognitive biases to test whether associations between SA and social cognition are moderated by such attentional and/or interpretation biases. Evidence of such moderation would suggest that social cognition in socially anxious individuals may be improved by incorporating attention training toward positive stimuli (Heeren et al., 2012; Li et al., 2008), or positive affect treatment (Craske et al., 2019), within cognitive behavioral therapy. Results from our review also found no apparent trend in findings of studies that included relevant covariates (age, gender, depressive symptoms, alexithymia, and IQ). Studies that included these covariates had mixed findings with effect sizes ranging from small to large. These results were comparable to studies that did not include covariates, suggesting that the variability in findings to date cannot be explained by the inclusion of covariates alone. Studies that included gender as a moderator found similar results (i.e., a positive association between SA and social cognition for females), which is supported by previous work demonstrating increased social cognitive ability in women (Babchuk et al., 1985; Thayer & Johnsen, 2000). However, only two studies examined this moderation, so future studies are needed to replicate this across domains of social cognition. Importantly, when examining specific domains of social cognition, alexithymia and depression seem to be relevant to consider when assessing affect sharing. Indeed, alexithymia is highly correlated with SA (Alvi et al., in press) and may mediate the association between SA and affect sharing (Morrison et al., 2016). Given the instructions in the affect sharing task that were used in the studies reviewed (i.e., assessing how you as a perceiver feel when watching a target recount an autobiographical story), it is likely that difficulty identifying emotions may impact performance on this task. Similarly, depression is highly comorbid with SA (Kessler et al., 2005) and shares an affective profile with SA of high negative and low positive affect (Kashdan, 2002), highlighting the importance of controlling for one when examining the other. As with studies that examined gender moderation, future work is needed to replicate the few findings of SA with affect sharing with a particular emphasis on these variables. Although findings based on the type of social cognitive assessment within each domain were mixed, there were some noticeable trends. For example, although only two studies included emotion recognition tasks involving vocal/prosody stimuli, both found a negative association with SA (Bodner et al., 2012; Quadflieg et al., 2007). Studies measuring theory of mind with the Reading the Mind in the Eyes Test (RMET; Baron-Cohen et al., 2001) were also generally consistent in their association with SA (71.4% found a statistically significant negative association). This trend suggests that the construct measured by this task, which has been conceptualized as theory of mind (Baron-Cohen et al., 2001), but also emotion recognition (Oakley et al., 2016), is reliably associated with symptoms of SA. Although this task may be more closely associated with emotion recognition, it differs from most lower-level emotion recognition tasks that rely on basic emotions (e.g., happy, sad, angry) and instead assesses complex emotions (e.g., jealous, cautious, grateful) with a restricted amount of information from facial expressions (i.e., only the eye region is presented). Since the RMET likely lies within a grey area between lower (i.e., emotion recognition) and higher-level social cognition (i.e., theory of mind), it may help to characterize this task as an intermediate-level social cognitive process moving forward. Alternatively, it could be characterized as theory of mind but distinguished by its assessment of decoding ability, rather than reasoning, due to the lack of context or additional information provided with the images (Sabbagh, 2004). Importantly, when examining studies based on the use of a decoding task (i.e., the RMET) vs. a reasoning task (i.e., MASC), results to date are still comparable. Most studies utilizing the MASC also found a significant negative association (i.e., 60% of studies), suggesting separation of findings based on decoding and reasoning tasks may not change the pattern of results. However, since far more studies of SA and social cognition have included the RMET compared to the MASC, this pattern of results should be viewed as preliminary. In sum, although most studies examining what we initially categorized as higher-level social cognitive processes found an association with SA, this finding is largely driven by studies that utilized the Reading the Mind in the Eyes Test. Thus, future within-person studies are needed to examine differential effects of SA on theory of mind that involves decoding vs. reasoning. 4.1 Limitations and future directions Our systematic review has several limitations that are important to point out. First, all aspects of social cognition were not included. This review focused on processes and tasks that are based on accuracy (or the correlation between perceiver and target responses as in the affect sharing task), however previous research has included domains such as attributional and attentional biases within social cognition (Green et al., 2012; Pinkham et al., 2016). Future reviews should examine findings in these additional areas to further understand the effect of SA on social cognition. Additionally, this review did not include unpublished studies, which may bias findings. Although unpublished studies may be of lesser quality due to the absence of the peer review process (Egger et al., 2003), publication bias may result from the exclusion of important findings in unpublished work. Indeed, unpublished studies often have smaller effects than published work (Hopewell et al., 2007); thus, the exclusion of grey literature may artificially inflate the overall significance and the effects of findings in the current review. Further, although this review focused on adult samples, SA may be related to social cognitive impairments in children and adolescents as well (Pittelkow et al., 2021). Future reviews should include studies in children and adolescents to determine if the trends in the present findings based on social cognitive domains, assessments, stimuli valence, and the inclusion of relevant covariates mirror those in studies with adult samples. Our review also identified that research on SA and social cognition is far less common in some domains (e.g., affect sharing) compared to others (e.g., emotion recognition). Future studies should focus on more understudied areas to examine whether effects from previous studies are replicated. In addition, given the heterogeneity among measures within each social cognitive domain, future studies examining SA and social cognition should include large samples with well-powered designs involving multiple measures of the same social cognitive domain to confirm that associations are domain rather than task-specific. Since behavioral tasks assessing social cognition are known to have poor psychometric properties (Pinkham et al., 2014), future research would also benefit from the inclusion of more well-validated tasks, such as the Geneva Emotion Recognition Test (Schlegel et al., 2019). Although our review did not identify differences based on type of SA assessment, it is difficult to interpret these findings based on only a few studies that have measured SA dimensionally. Symptoms of psychological distress are thought to be more accurately operationalized using dimensional measures (Kessler, 2002), and creating groups based on cutoffs and relying on diagnostic groups may restrict the range of symptoms. Thus, future studies may further elucidate the association of SA and social cognition if SA is measured dimensionally. Finally, future research examining the association between SA and social cognition would also benefit from the assessment of real-world social functioning. Although it is typically assumed that laboratory-based assessments of social cognitive performance correlate with real world functioning, there is limited work in this area as it pertains to SA. Studies employing daily diary, or ecological momentary assessment (Hur et al., 2019), would be particularly beneficial in this regard as this would allow researchers to extend their findings beyond the laboratory to examine whether differences in social cognition among socially anxious individuals translate into difficulties outside of the laboratory. 5 Conclusion and implications In sum, this systematic review highlights several important findings. First, although there is mixed evidence for the association between SA and lower-level social cognitive processes, there is a trend in studies showing negative associations between SA and intermediate to higher-level social cognitive processes (theory of mind and empathic accuracy). Second, findings across studies suggest that stimuli valence is important to consider as a moderator in these associations. Of the studies that examined valence-specific associations, most found a significant negative association between SA and the ability to accurately perceive positive and neutral stimuli. Third, this review highlights the lack of inclusion of statistical covariates across studies, which may be most relevant when examining SA and affect sharing. To further clarify the relation between SA and social cognition, future studies would benefit from the inclusion of relevant moderators and covariates, multiple measures within the same domain of social cognition, and dimensional assessment of SA. Findings related to impairment in higher-level social cognitive processes among those with SA symptoms may be explained cognitive models of SA, which propose that social cognitive misinterpretations result in and maintain symptoms (Beck, 2011; Clark and Wells, 1995). For example, individuals with SA may misinterpret social cues in everyday life resulting in automatic negative thoughts. These negative interpretations impact social behaviour (e.g., through avoidance of social situations) and interpersonal functioning. Thus, impairment in higher-level social cognitive processes, which includes mental state inference of complex social stimuli, is consistent with this theoretical understanding of SA. Likewise, this aligns with the cognitive model of SA which posits that emotions and behaviors are influenced by individuals' thoughts (Beck, 2011). Thus, these findings highlight the utility of focusing on cognitive, as opposed to emotional, processes in psychotherapy for SAD. In addition, based on the present findings, it may be more beneficial for clinicians to focus on higher-level, rather than lower-level social cognitive processes. Indeed, our findings suggest that trials examining the efficacy of social skills training in combination with cognitive behavioral therapy (Beidel et al., 2014) may benefit from even more targeted social cognitive skills training (Kurtz et al., 2016). Future research using social cognitive interventions in individuals with SA is needed to examine these potential clinical implications. CRediT authorship contribution statement Talha Alvi: Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Project Administration, Writing-original draft, Writing-reviewing and editing. Divya Kumar: Formal analysis, Project administration, Writing-review & editing. Benjamin A. Tabak: Conceptualization, Supervision, Writing-review & editing. Role of the funding source This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors. Conflict of interest All authors declare no conflicts of interest. Acknowledgements We thank Dr. Austin Baldwin for providing helpful feedback on an earlier draft of this manuscript. ==== Refs References Aderka I.M. Hofmann S.G. Nickerson A. Hermesh H. Gilboa-Schechtman E. Marom S. Functional impairment in social anxiety disorder J. Anxiety Disord. 26 3 2012 393 400 10.1016/J.JANXDIS.2012.01.003 22306132 Arrais K.C. Machado-de-Sousa J.P. Trzesniak C. Filho A.S. Ferrari M.C.F. Osório F.L. Loureiro S.R. Nardi A.E. Hetem L.A.B. Zuardi A.W. Hallak J.E.C. Crippa J.A.S. Social anxiety disorder women easily recognize fearfull, sad and happy faces: the influence of gender J. Psychiatr. Res. 44 8 2010 535 540 10.1016/j.jpsychires.2009.11.003 19962717 Alden L.E. Taylor C.T. Interpersonal processes in social phobia Clin. Psychol. Rev. 24 7 2004 857 882 10.1016/j.cpr.2004.07.006 15501559 Alvi T. Kouros C.D. Lee J. Fulford D. Tabak B.A. Social anxiety is negatively associated with theory of mind and empathic accuracy J. Abnorm. Psychol. 129 1 2020 10.1037/abn0000493 Alvi T. Rosenfield D. Sunahara C.S. Wallmark Z. Lee J. Tabak B.A. Examining unique associations of social anxiety and depression on behaviorally assessed affective empathy Clin. Psychol. Sci. 2022 in press American Psychiatric Association DSM V Diagnostic and Statistical Manual of Mental Disorders Fifth edition 2013 10.1109/mwc.2007.4300985 Auyeung K. Alden L.E. Accurate empathy, social rejection, and social anxiety disorder Clin. Psychol. Sci. 8 2 2020 266 279 10.1177/2167702619885410 Auyeung K. Alden L.E. Social anxiety and empathy for social pain Cogn. Ther. Res. 2016 10.1007/s10608-015-9718-0 Babchuk W.A. Hames R.B. Thompson R.A. Sex differences in the recognition of infant facial expressions of emotion: the primary caretaker hypothesis Ethol. Sociobiol. 6 2 1985 89 101 10.1016/0162-3095(85)90002-0 Ballespí S. Pérez-Domingo A. Vives J. Sharp C. Barrantes-Vidal N. Childhood behavioral inhibition is associated with impaired mentalizing in adolescence PLoS ONE 13 3 2018 10.1371/journal.pone.0195303 Banerjee M. Hidden Emotions: Preschoolers’ Knowledge of Appearance-reality and Emotion Display Rules Soc. Cogn. 15 2 1997 107 132 10.1521/SOCO.1997.15.2.107 Doi:10.1521/Soco.1997.15.2.107 Baron-Cohen S. Mindblindness: An essay on autism and theory of mind Mindblindness: An Essay on Autism and Theory of Mind 1995 The MIT Press Baron-Cohen S. O'Riordan M. Stone V. Jones R. Plaisted K. Recognition of faux pas by normally developing children and children with asperger syndrome or high-functioning autism J. Autism Dev. Disord. 29 5 1999 407 418 10.1023/a:1023035012436 10587887 Baron-Cohen S. Wheelwright S. Hill J. Raste Y. Plumb I. The “Reading the mind in the eyes” test revised version : a study with normal adults, and adults with asperger syndrome or high-functioning autism J. Child Psychol. Psychiatry 42 2 2001 241 251 10.1111/1469-7610.00715 11280420 Beck J. Cognitive Behavior Therapy: Basics and Beyond 2011 The Guilford Press 10.1017/CBO9781107415324.004 Bell C. Bourke C. Colhoun H. Carter F. Frampton C. Porter R. The misclassification of facial expressions in generalised social phobia J. Anxiety Disord. 25 2 2011 278 283 10.1016/j.janxdis.2010.10.001 21041060 Beidel D.C. Alfano C.A. Kofler M.J. Rao P.A. Scharfstein L. Wong Sarver N. The impact of social skills training for social anxiety disorder: a randomized controlled trial J. Anxiety Disord. 28 8 2014 908 918 10.1016/J.JANXDIS.2014.09.016 25445081 Berger Y. Ability to Express and Perceive Emotional Vocal Expressions, by Verbal and Non-verbal Learning Disabled Children and by Children Without Learning Disabilities 2002 Ilan University Ramat Gan MA Thesis Bodner E. Aharoni R. Iancu I. The effect of training with music on happiness recognition in social anxiety disorder J. Psychopathol. Behav. Assess. 34 4 2012 458 466 10.1007/s10862-012-9304-7 Bodner E. Gilboa A. Emotional communicability in music therapy: different instruments for different emotions? Nord. J. Music. Ther. 15 1 2006 3 16 10.1080/08098130609478147 Bora E. Berk M. Theory of mind in major depressive disorder: a meta-analysis J. Affect. Disord. 191 2016 49 55 10.1016/j.jad.2015.11.023 26655114 Brothers L. The neural basis of primate social communication Motiv. Emot. 14 2 1990 81 91 10.1007/BF00991637 Brüne M. Theory of mind and the role of IQ in chronic disorganized schizophrenia Schizophr. Res. 60 1 2003 57 64 10.1016/S0920-9964(02)00162-7 12505138 Buhlmann U. Wacker R. Dziobek I. Inferring other people’s states of mind: comparison across social anxiety, body dysmorphic, and obsessive-compulsive disorders J. Anxiety Disord. 34 2015 107 113 10.1016/j.janxdis.2015.06.003 26218178 Button K. Lewis G. Penton-Voak I. Munafò M. Social anxiety is associated with general but not specific biases in emotion recognition Psychiatry Res. 210 1 2013 199 207 10.1016/j.psychres.2013.06.005 23845415 Campbell D.W. Sareen J. Stein M.B. Kravetsky L.B. Paulus M.P. Hassard S.T. Reiss J.P. Happy but not so approachable: the social judgments of individuals with generalized social phobia Depress. Anxiety 26 5 2009 419 424 10.1002/da.20474 19242987 Chen S.H. Emotion Recognition and Social Rehabilitation in Patients with Schizophrenia: Development and Revision of Measure for Nonverbal Emotion Recognition (DOH94-TD-M-113-020; DOH95-TD-M-113-006) 2006 Department of Health, Science and Technology Unit Clark D.A. Wells A. A cognitive model of social phobia Social Phobia: Diagnosis, Assessment, and Treatment 1995 Conzelmann K. Weis S. Süß H.-M. New findings about social intelligence J. Individ. Differ. 34 3 2013 119 137 10.1027/1614-0001/a000106 Costanzo M. Archer D. Interpreting the expressive behavior of others: the interpersonal perception task J. Nonverbal Behav. 13 1989 225 245 10.1007/BF00990295 Craske M.G. Meuret A.E. Ritz T. Treanor M. Dour H. Rosenfield D. Positive affect treatment for depression and anxiety: a randomized clinical trial for a core feature of anhedonia J. Consult. Clin. Psychol. 87 5 2019 457 471 10.1037/ccp0000396 30998048 Davila J. Beck J.G. Is social anxiety associated with impairment in close relationships?A preliminary investigation Behavior Therapy 33 3 2002 427 446 10.1016/S0005-7894(02)80037-5 Davis M.H. Measuring individual differences in empathy: evidence for a multidimensional approach J. Pers. Soc. Psychol. 1983 10.1037/0022-3514.44.1.113 DePaulo B.M. Decoding discrepant nonverbal cues J. Pers. Soc. Psychol. 36 3 1978 313 323 10.1037/0022-3514.36.3.313 Di Tella Marialaura Adenzato Mauro Catmur Caroline Miti Francesca Castelli Lorys Ardito Rita The role of alexithymia in social cognition: Evidence from a non-clinical population Journal of Affective Disorders 273 2020 482 492 10.1016/j.jad.2020.05.012 32560944 Dickter C.L. Burk J.A. Fleckenstein K. Kozikowski C.T. Autistic traits and social anxiety predict differential performance on social cognitive tasks in typically developing young adults PLoS ONE 13 3 2018 10.1371/journal.pone.0195239 Doherty R.W. Orimoto L. Singelis T.M. Hatfield E. Hebb J. Emotional contagion: gender and occupational differences Psychol. Women Q. 19 1995 355 371 10.1111/j.1471-6402.1995.tb00080.x Dziobek I. Fleck S. Kalbe E. Rogers K. Hassenstab J. Brand M. Kessler J. Woike J.K. Wolf O.T. Convit A. Introducing MASC: a movie for the assessment of social cognition J. Autism Dev. Disord. 2006 10.1007/s10803-006-0107-0 Ebner N.C. Riediger M. Lindenberger U. FACES–a database of facial expressions in young, middle-aged, and older women and men: development and validation Behav. Res. Methods 42 1 2010 351 362 10.3758/BRM.42.1.351 20160315 Egger M. Juni P. Bartlett C. Holenstein F. Sterne J. How important are comprehensive literature searches and the assessment of trial quality in systematic reviews? Empirical study Health Technol. Assess. (Winch. Eng.) 7 1 2003 1 76 Ekman P. Friesen W.V. Pictures of Facial Affect 1976 Consulting Psychologists Press Palo Alto, CA Fan Y. Han S. Temporal dynamic of neural mechanisms involved in empathy for pain: an event-related brain potential study Neuropsychologia 46 1 2008 160 173 10.1016/j.neuropsychologia.2007.07.023 17825852 Farmer A.S. Kashdan T.B. Social anxiety and emotion regulation in daily life: spillover effects on positive and negative social events Cogn. Behav. Ther. 2012 10.1080/16506073.2012.666561 Frith C.D. Frith U. Social cognition in humans Current Biology 2007 10.1016/j.cub.2007.05.068 Garner M. Baldwin D.S. Bradley B.P. Mogg K. Impaired identification of fearful faces in generalised social phobia J. Affect. Disord. 115 3 2009 460 465 10.1016/j.jad.2008.10.020 19062103 Gilboa-Schechtman E. Foa E. Vaknin Y. Marom S. Hermesh H. Interpersonal sensitivity and response bias in social phobia and depression: labeling emotional expressions Cogn. Ther. Res. 32 5 2008 605 618 10.1007/s10608-008-9208-8 Gleason K.A. Jensen-Campbell L.A. Ickes W. The role of empathic accuracy in adolescents’ peer relations and adjustment Personal. Soc. Psychol. Bull. 35 8 2009 997 1011 10.1177/0146167209336605 Green M.F. Bearden C.E. Cannon T.D. Fiske A.P. Hellemann G.S. Horan W.P. Kee K. Kern R.S. Lee J. Sergi M.J. Subotnik K.L. Sugar C.A. Ventura J. Yee C.M. Nuechterlein K.H. Social cognition in schizophrenia, part 1: performance across phase of illness Schizophr. Bull. 2012 10.1093/schbul/sbq171 Green M.F. Lee J. Ochsner K.N. Adapting social neuroscience measures for schizophrenia clinical trials, part 1: ferrying paradigms across perilous waters Schizophr. Bull. 2013 10.1093/schbul/sbt131 Green M.F. Penn D.L. Bentall R. Carpenter W.T. Gaebel W. Gur R.C. Kring A.M. Park S. Silverstein S.M. Heinssen R. Social cognition in schizophrenia: an NIMH workshop on definitions, assessment, and research opportunities Schizophr. Bull. 2008 10.1093/schbul/sbm145 Gur R.C. Sara R. Hagendoorn M. Marom O. Hughett P. Macy L. Turner T. Bajcsy R. Posner A. Gur R.E. A method for obtaining 3-dimensional facial expressions and its standardization for use in neurocognitive studies J. Neurosci. Methods 115 2 2002 137 143 10.1016/s0165-0270(02)00006-7 11992665 Hagemann J. Straube T. Schulz C. Too bad: bias for angry faces in social anxiety interferes with identity processing Neuropsychologia 84 2016 136 149 10.1016/j.neuropsychologia.2016.02.005 26878979 Halberstadt J.B. Niedenthal P.M. Emotional state and the use of stimulus dimensions in judgment J. Pers. Soc. Psychol. 72 5 1997 1017 1033 10.1037//0022-3514.72.5.1017 9150582 Hampel S. Weis S. Hiller W. Witthöft M. The relations between social anxiety and social intelligence: a latent variable analysis J. Anxiety Disord. 25 4 2011 545 553 10.1016/j.janxdis.2011.01.001 21315550 Hatfield E. Cacioppo J.T. Rapson R.L. Primitive emotional contagion Emotion and Social Behavior 1992 Hawes M.T. Szenczy A.K. Klein D.N. Hajcak G. Nelson B.D. Increases in depression and anxiety symptoms in adolescents and young adults during the COVID-19 pandemic Psychol. Med. 2021 10.1017/S0033291720005358 Heberlein A.S. Adolphs R. Tranel D. Damasio H. Cortical regions for judgments of emotions and personality traits from point-light walkers J. Cogn. Neurosci. 16 7 2004 1143 1158 10.1162/0898929041920423 15453970 Heeren A. Reese H.E. McNally R.J. Philippot P. Attention training toward and away from threat in social phobia: effects on subjective, behavioral, and physiological measures of anxiety Behav. Res. Ther. 50 1 2012 30 39 10.1016/j.brat.2011.10.005 22055280 Hess U. Blairy S. Set of Emotional Facial Stimuli 1995 Department of Psychology, University of Quebec at Montreal Montreal, Canada Heuer K. Lange W.-G. Isaac L. Rinck M. Becker E.S. Morphed emotional faces: emotion detection and misinterpretation in social anxiety J. Behav. Ther. Exp. Psychiatry 41 4 2010 418 425 10.1016/j.jbtep.2010.04.005 20511123 Hezel D.M. McNally R.J. Theory of mind impairments in social anxiety disorder Behav. Ther. 45 4 2014 530 540 10.1016/j.beth.2014.02.010 24912465 Holt-Lunstad J. The major health implications of social connection Current Directions in Psychology Science 30 3 2021 10.1177/2F0963721421999630 Hopewell S. McDonald S. Clarke M. Egger M. Grey literature in meta-analyses of randomized trials of health care interventions Cochrane Database Syst. Rev. 2 2007 MR000010 10.1002/14651858.MR000010.pub3 Horstmann G. Facial expressions of emotion: does the prototype represent central tendency, frequency of instantiation, or an ideal? Emotion 2 3 2002 297 305 10.1037/1528-3542.2.3.297 12899362 Hunter L.R. Buckner J.D. Schmidt N.B. Interpreting facial expressions: the influence of social anxiety, emotional valence, and race J. Anxiety Disord. 23 4 2009 482 488 10.1016/j.janxdis.2008.10.004 19026520 Hur J. DeYoung K.A. Islam S. Anderson A.S. Barstead M.G. Shackman A.J. Social context and the real-world consequences of social anxiety Psychol. Med. 50 12 2019 1989 2000 10.1017/S0033291719002022 31423954 Ickes W. Stinson L. Bissonnette V. Garcia S. Naturalistic social cognition: empathic accuracy in mixed-sex dyads J. Pers. Soc. Psychol. 1990 10.1037/0022-3514.59.4.730 Jacobs M. Snow J. Geraci M. Vythilingam M. Blair R.J.R. Charney D.S. Pine D.S. Blair K.S. Association between level of emotional intelligence and severity of anxiety in generalized social phobia J. Anxiety Disord. 22 8 2008 1487 1495 10.1016/j.janxdis.2008.03.003 18439799 Joormann J. Gotlib I.H. Is this happiness i see? Biases in the identification of emotional facial expressions in depression and social phobia J. Abnorm. Psychol. 115 4 2006 705 714 10.1037/0021-843X.115.4.705 17100528 Kang S.M. Lau A.S. Revisiting the out-group advantage in emotion recognition in a multicultural society: further evidence for the in-group advantage Emotion 13 2 2013 203 215 10.1037/a0030013 23163711 Kashdan T.B. Social anxiety dimensions, neuroticism, and the contours of positive psychological functioning Cogn. Ther. Res. 2002 10.1023/A:1021293501345 Kashdan T.B. Volkmann J.R. Breen W.E. Han S. Social anxiety and romantic relationships: the costs and benefits of negative emotion expression are context-dependent J. Anxiety Disord. 21 4 2007 475 492 10.1016/J.JANXDIS.2006.08.007 17045778 Kern R.S. Penn D.L. Lee J. Horan W.P. Reise S.P. Ochsner K.N. Marder S.R. Green M.F. Adapting social neuroscience measures for schizophrenia clinical trials, part 2: trolling the depths of psychometric properties Schizophr. Bull. 39 6 2013 1201 1210 10.1093/schbul/sbt127 24072805 Kessler R.C. The categorical versus dimensional assessment controversy in the sociology of mental illness J. Health Soc. Behav. 43 2 2002 171 10.2307/3090195 12096698 Kessler R.C. Berglund P. Demler O. Jin R. Merikangas K.R. Walters E.E. Lifetime prevalence and age-of-onset distributions of DSM-IV disorders in the national comorbidity survey replication Arch. Gen. Psychiatry 2005 10.1001/archpsyc.62.6.593 Kurtz M.M. Gagen E. Rocha N.B.F. Machado S. Penn D.L. Comprehensive treatments for social cognitive deficits in schizophrenia: a critical review and effect-size analysis of controlled studies Clin. Psychol. Rev. 43 2016 80 89 10.1016/J.CPR.2015.09.003 26437567 Langner O. Dotsch R. Bijlstra G. Wigboldus D.H.J. Hawk S.T. van Knippenberg A. Presentation and validation of the radboud faces database Cognit. Emot. 24 8 2010 1377 1388 10.1080/02699930903485076 Lau A.S. Wang S. Fung J.J. Namikoshi M. What happens when you “can’t read the air”? Cultural fit and aptitude by values interactions on social anxiety J. Soc. Clin. Psychol. 33 10 2014 853 866 10.1521/jscp.2014.33.10.853 Lenton-Brym A.P. Moscovitch D.A. Vidovic V. Nilsen E. Friedman O. Theory of mind ability in high socially anxious individuals Anxiety Stress Coping 31 5 2018 487 499 http://10.0.4.56/10615806.2018.1483021 29940803 Li S. Tan J. Qian M. Liu X. Continual training of attentional bias in social anxiety Behav. Res. Ther. 46 8 2008 905 912 10.1016/j.brat.2008.04.005 18538305 Lundqvist D. Flykt A. Öhman A. The Karolinska Directed Emotional Faces - KDEF, CD ROM from Department of Clinical Neuroscience 1998 Psychology section, Karolinska Institutet 91-630-7164-9 Lyvers M. Scott K. Thorberg F.A. Social anxiety and alexithymia in relation to problematic drinking and theory of mind Am. J. Psychol. 132 3 2019 325 341 10.5406/amerjpsyc.132.3.0325 Magen E. Konasewich P.A. Women support providers are more susceptible than men to emotional contagion following brief supportive interactions Psychol. Women Q. 35 4 2011 611 616 10.1177/0361684311423912 Maleki G. Zabihzadeh A. Richman M.J. Demetrovics Z. Mohammadnejad F. Decoding and reasoning mental states in major depression and social anxiety disorder BMC Psychiatry 20 2020 10.1186/s12888-020-02873-w Matsumoto D. Ekman P. Japanese and Caucasian Facial Expressions of Emotion (JACFEE) [Slides] 1988 Intercultural and Emotion Research Laboratory, Department of Psychology, San Francisco State University San Francisco, CA Mayer J.D. Salovey P. Caruso D. Mayer-salovey-caruso emotional intelligence test (MSCEIT) users manual 2002 Multi-Health Systems Toronto, Ontario McDonald S. Bornhofen C. Shum D. Long E. Saunders C. Neulinger K. Reliability and validity of the awareness of social inference test (TASIT): a clinical test of social perception Disabil. Rehabil. 28 24 2006 1529 1542 10.1080/09638280600646185 17178616 Mohlman J. Carmin C.N. Price R.B. Jumping to interpretations: social anxiety disorder and the identification of emotional facial expressions Behav. Res. Ther. 45 3 2007 591 599 10.1016/j.brat.2006.03.007 16643844 Moher D. Shamseer L. Clarke M. Ghersi D. Liberati A. Petticrew M. Preferred reporting items for systematic review and meta-analysis protocols (PRISMA-P) 2015 statement Syst. Rev. 4 1 2015 Montagne B. Kessels R.P.C. De Haan E.H.F. Perrett D.I. The emotion recognition task: a paradigm to measure the perception of facial emotional expressions at different intensities Percept. Mot. Skills 104 2 2007 589 598 10.2466/pms.104.2.589-598 17566449 Montagne B. Schutters S. Westenberg H.G.M. van Honk J. Kessels R.P.C. de Haan E.H.F. Reduced sensitivity in the recognition of anger and disgust in social anxiety disorder Cognitive Neuropsychiatry 11 4 2006 389 401 10.1080/13546800444000254 17354077 Morrison A.S. Heimberg R.G. Social anxiety and social anxiety disorder Annu. Rev. Clin. Psychol. 9 1 2013 249 274 10.1146/annurev-clinpsy-050212-185631 23537485 Morrison A.S. Mateen M.A. Brozovich F.A. Zaki J. Goldin P.R. Heimberg R.G. Gross J.J. Empathy for positive and negative emotions in social anxiety disorder Behav. Res. Ther. 87 2016 232 242 10.1016/j.brat.2016.10.005 27816799 Mullins D.T. Duke M.P. Effects of social anxiety on nonverbal accuracy and response time I: facial expressions J. Nonverbal Behav. 28 1 2004 3 33 10.1023/B:JONB.0000017865.24656.98 Murphy B.A. Lilienfeld S.O. Are self-report cognitive empathy ratings valid proxies for cognitive empathy ability? Negligible meta-analytic relations with behavioral task performance Psychol. Assess. 31 8 2019 1062 1072 10.1037/pas0000732 31120296 Nowicki S. Carton J. The measurement of emotional intensity from facial expressions J. Soc. Psychol. 1993 10.1080/00224545.1993.9713934 Nowicki S. Jr. Glanville D. Demertzis A. A test of the ability to recognize emotion in the facial expressions of african american adults J. Black Psychol. 24 3 1998 335 350 10.1177/00957984980243005 Oh K.-S. Lee W.H. Kim S. Shin D.-W. Shin Y.-C. Lim S.-W. Impaired facial expression recognition in patients with social anxiety disorder: a case-control study Cogn. Neuropsychiatry 23 4 2018 218 228 10.1080/13546805.2018.1462695 29633666 O’Toole M.S. Hougaard E. Mennin D.S. Social anxiety and emotion knowledge: a meta-analysis J. Anxiety Disord. 2013 10.1016/j.janxdis.2012.09.005 Oakley B.F.M. Brewer R. Bird G. Catmur C. Theory of mind is not theory of emotion: a cautionary note on the reading the mind in the eyes test J. Abnorm. Psychol. 125 6 2016 818 823 10.1037/abn0000182 27505409 Penn D.L. Ritchie M. Francis J. Combs D. Martin J. Social perception in schizophrenia: the role of context Psychiatry Res. 109 2 2002 149 159 10.1016/S0165-1781(02)00004-5 11927140 Phan K.L. Fitzgerald D.A. Nathan P.J. Tancer M.E. Association between amygdala hyperactivity to harsh faces and severity of social anxiety in generalized social phobia Biol. Psychiatry 59 5 2006 424 429 10.1016/j.biopsych.2005.08.012 16256956 Philippot P. Douilliez C. Social phobics do not misinterpret facial expression of emotion Behav. Res. Ther. 43 5 2005 639 652 10.1016/j.brat.2004.05.005 15865918 Pinkham A.E. Penn D.L. Green M.F. Buck B. Healey K. Harvey P.D. The social cognition psychometric evaluation study: results of the expert survey and RAND panel Schizophr. Bull. 40 4 2014 813 823 10.1093/schbul/sbt081 23728248 Pinkham A.E. Penn D.L. Green M.F. Harvey P.D. Social cognition psychometric evaluation: results of the initial psychometric study Schizophr. Bull. 2016 10.1093/schbul/sbv056 Pittelkow M.-M. Aan Het Rot M. Seidel L.J. Feyel N. Roest A.M. Social anxiety and empathy: a systematic review and meta-analysis Journal of Anxiety Disorders 78 2021 102357 10.1016/j.janxdis.2021.102357 Plana I. Lavoie M.-A. Battaglia M. Achim A.M. A meta-analysis and scoping review of social cognition performance in social phobia, posttraumatic stress disorder and other anxiety disorders J. Anxiety Disord. 28 2 2014 169 177 10.1016/j.janxdis.2013.09.005 24239443 Quadflieg S. Wendt B. Mohr A. Miltner W.H.R. Straube T. Recognition and evaluation of emotional prosody in individuals with generalized social phobia: a pilot study Behav. Res. Ther. 45 12 2007 3096 3103 10.1016/j.brat.2007.08.003 17880917 Ruffman T. Henry J.D. Livingstone V. Phillips L.H. A meta-analytic review of emotion recognition and aging: implications for neuropsychological models of aging Neurosci. Biobehav. Rev. 32 4 2008 863 881 10.1016/j.neubiorev.2008.01.001 18276008 Sabbagh M.A. Understanding orbitofrontal contributions to theory-of-mind reasoning: implications for autism Brain Cogn. 55 1 2004 209 219 10.1016/j.bandc.2003.04.002 15134854 Schlegel K. Fontaine J.R.J. Scherer K.R. The nomological network of emotion recognition ability Eur. J. Psychol. Assess. 35 3 2019 352 363 10.1027/1015-5759/a000396 Schneier F.R. Heckelman L.R. Garfinkel R. Campeas R. Fallon B.A. Gitow A. Street L. Del Bene D. Liebowitz M.R. Functional impairment in social phobia J. Clin. Psychiatry 55 8 1994 322 331 8071299 Schofield C.A. Coles M.E. Gibb B.E. Social anxiety and interpretation biases for facial displays of emotion: Emotion detection and ratings of social cost Behav. Res. Ther. 45 12 2007 2950 2963 10.1016/j.brat.2007.08.006 17868640 Schroeder J.E. Interpersonal perception skills: self-concept correlates Percept. Mot. Skills 80 1 1995 51 56 10.2466/pms.1995.80.1.51 Schroeder J.E. Self-concept, social anxiety, and interpersonal perception skills Personal. Individ. Differ. 19 6 1995 955 958 10.1016/S0191-8869(95)00108-5 Schroeder J.E. Ketrow S.M. Social anxiety and performance in an interpersonal perception task Psychological Reports 81 3, Pt 1 1997 991 996 10.2466/pr0.1997.81.3.991 9400088 Sened H. Lavidor M. Lazarus G. Bar-Kalifa E. Rafaeli E. Ickes W. Empathic accuracy and relationship satisfaction: a meta-analytic review J. Fam. Psychol. 31 6 2017 742 752 10.1037/FAM0000320 28394141 Singer T. Lamm C. The social neuroscience of empathy Ann. N. Y. Acad. Sci. 2009 10.1111/j.1749-6632.2009.04418.x Straube T. Kolassa I.-T. Glauer M. Mentzel H.-J. Miltner W.H.R. Effect of task conditions on brain responses to threatening faces in social phobics: an event-related functional magnetic resonance imaging study Biol. Psychiatry 56 12 2004 921 930 10.1016/j.biopsych.2004.09.024 15601601 Sutterby S.R. Bedwell J.S. Passler J.S. Deptula A.E. Mesa F. Social anxiety and social cognition: the influence of sex Psychiatry Res. 197 3 2012 242 245 10.1016/j.psychres.2012.02.014 22424909 Taylor C.T. Bomyea J. Amir N. Attentional bias away from positive social information mediates the link between social anxiety and anxiety vulnerability to a social stressor J. Anxiety Disord. 2010 10.1016/j.janxdis.2010.02.004 Thayer J.F. Johnsen B.H. Sex differences in judgement of facial affect: a multivariate analysis of recognition errors Scand. J. Psychol. 41 3 2000 243 246 10.1111/1467-9450.00193 11041306 Tibi-Elhanany Y. Shamay-Tsoory S.G. Social cognition in social anxiety: first evidence for increased empathic abilities Isr. J. Psychiatry Relat. Sci. 48 2 2011 98 106 22120444 Tibi-Elhanany Y. Shamay-Tsoory S.G. Shamay-Tsoory S. Social cognition in social anxiety: first evidence for increased empathic abilities Isr. J. Psychiatry Relat. Sci. 48 2 2011 Tonge N.A. Lim M.H. Piccirillo M.L. Fernandez K.C. Langer J.K. Rodebaugh T.L. Interpersonal problems in social anxiety disorder across different relational contexts J. Anxiety Disord. 75 2020 102275 10.1016/J.JANXDIS.2020.102275 Torro-Alves N. Claudino R.G.E. Rodrigues M.R. Machado-de-Sousa J.P. Crippa J.A. Bezerra I.A.de O. Osório F.de L. Facial emotion recognition in social anxiety: the influence of dynamic information Psychology & Neuroscience 9 1 2016 1 11 10.1037/pne0000042 Tottenham N. Tanaka J.W. Leon A.C. McCarry T. Nurse M. Hare T.A. Marcus D.J. Westerlund A. Casey B.J. Nelson C. The NimStim set of facial expressions: judgments from untrained research participants Psychiatry Res. 168 3 2009 242 249 10.1016/j.psychres.2008.05.006 19564050 Tseng H.-H. Huang Y.-L. Chen J.-T. Liang K.-Y. Lin C.-C. Chen S.-H. Facial and prosodic emotion recognition in social anxiety disorder Cogn. Neuropsychiatry 22 4 2017 331 345 10.1080/13546805.2017.1330190 28537109 Veljaca K.-A. Rapee R.M. Detection of negative and positive audience behaviours by socially anxious subjects Behav. Res. Ther. 36 3 1998 311 321 10.1016/S0005-7967(98)00016-3 9642850 Washburn D. Wilson G. Roes M. Rnic K. Harkness K.L. Theory of mind in social anxiety disorder, depression, and comorbid conditions J. Anxiety Disord. 37 2016 71 77 10.1016/j.janxdis.2015.11.004 26658117 Wendt B. Scheich H. The ‘Magdeburger Prosodie-Korpus’ Bel B. Martien I. Proceedings of the Speech Prosody 2002 Conference, 11– 13 April 2002 2002 Laboratoire Parole et Langage Aix-en-Provence 699 701 in press Winton E.C. Clark D.M. Edelmann R.J. Social anxiety, fear of negative evaluation and the detection of negative emotion in others Behav. Res. Ther. 33 2 1995 193 196 10.1016/0005-7967(94)E0019-F 7887878 Wittchen H.U. Fuetsch M. Sonntag H. Müller N. Liebowitz M. Disability and quality of life in pure and comorbid social phobia. Findings from a controlled study Eur. Psychiatry 15 1 2000 46 58 10.1016/S0924-9338(00)00211-X 10713802 Yoon K.L. Fitzgerald D.A. Angstadt M. McCarron R.A. Phan K.L. Amygdala reactivity to emotional faces at high and low intensity in generalized social phobia: a 4-Tesla functional MRI study Psychiatry Res. 154 1 2007 93 98 10.1016/j.pscychresns.2006.05.004 17097275 Young A. Perrett D. Calder A. Sprengelmeyer R. Ekman P. Facial Expressions of Emotion – Stimuli and Tests (FEEST) 2002 Thames Valley Test Bury St. Edmunds, England Zaki J. Bolger N. Ochsner K. It takes two: the interpersonal nature of empathic accuracy Psychol. Sci. 19 2008 399 404 10.1111/j.1467-9280.2008.02099.x 18399894

35490878

PMC9754122

NO-CC CODE

2022-12-16 23:26:27

no

J Affect Disord. 2022 Aug 15; 311:17-30

utf-8

J Affect Disord

10.1016/j.jad.2022.04.130

oa_other

==== Front Comp Immunol Microbiol Infect Dis Comp Immunol Microbiol Infect Dis Comparative Immunology, Microbiology and Infectious Diseases 0147-9571 1878-1667 Elsevier Ltd. S0147-9571(22)00057-1 10.1016/j.cimid.2022.101800 101800 Letter to Editor Serosurvey for Nipah virus in bat population of southern part of India Gokhale Mangesh a1 Sudeep A.B. a1 Mathapati Basavaraj a1 Balasubramanian R. b Ullas P.T. a Mohandas Sreelekshmy a Patil Dilip R. a Shete Anita M. a Gopale Sanjay a Sawant Pradeep a Jain Rajlaxmi a Holeppanavar Manjunath a Suryawanshi Annasaheb T. a Chopade Ganesh a Dhaigude Sachin a Patil Deepak Y. a Mourya Devendra T. a Yadav Pragya D. a⁎ a Indian Council of Medical Research-National Institute of Virology, Pune 411021, India b Indian Council of Medical Research-National Institute of Virology, Field Unit, Vandanam, Kerala 688005, India ⁎ Correspondence to: Maximum Containment Facility, Indian Council of Medical Research, National Institute of Virology, Sus Road, Pashan, Pune 411021, India. 1 Equal first author. 26 3 2022 6 2022 26 3 2022 85 101800101800 28 12 2021 22 3 2022 23 3 2022 © 2022 Elsevier Ltd. All rights reserved. 2022 Elsevier Ltd Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. Nipah virus (NiV) is one of the priority pathogens with pandemic potential. Though the spread is far slower than SARS-CoV-2, case fatality is the biggest concern. Fruit bats belonging to genus Pteropus are identified to be the main reservoir of the virus causing sporadic cases and outbreaks in Malaysia, Bangladesh and India. The sudden emergence of Nipah in Kerala, India during 2018–2019 has been astonishing with respect to its introduction in the unaffected areas. With this, active Nipah virus surveillance was conducted among bat populations in Southern part of India viz., Karnataka, Kerala, Tamil Nadu, Telangana, Puducherry and Odisha during January-November 2019. Throat swabs/rectal swabs (n = 573) collected from Pteropus medius and Rousettus leschenaultii bat species and sera of Pteropus medius bats (n = 255) were screened to detect the presence of Nipah viral RNA and anti-Nipah IgG antibodies respectively. Of 255 P. medius bats sera samples, 51 bats (20%) captured from Karnataka, Kerala, Tamil Nadu and Puducherry demonstrated presence of anti-Nipah IgG antibodies. However, the presence of virus couldn’t be detected in any of the bat specimens. The recent emergence of Nipah virus in Kerala in September 2021 warrants further surveillance of Nipah virus among bat populations from the affected and remaining states of India. Keywords Nipah Bat Pteropus Surveillance South India ==== Body pmc1 Introduction Nipah virus (NiV) is an emerging paramyxovirus capable of causing lethal infection in a number of mammalian species including humans. The first human infection with NiV was identified during an outbreak of severe encephalitis in Malaysia in 1998–1999 [1]. Both animal-to-humans and human-to-human transmission has been documented during different outbreaks. More than 700 human cases of Nipah virus infections were reported from Malaysia, India, Bangladesh, Singapore and Philippines during 1998−2018 [2]. India has witnessed four outbreaks of Nipah virus disease during 2001–2019. The first outbreak of NiV with presentation of febrile illness and neurological symptoms was observed among human population in Siliguri, West Bengal during 2001 with a case fatality rate (CFR) of 74% [3]. Subsequently, an outbreak was reported from Nadia district, West Bengal in 2007 which affected five people and all succumbed to infection (CFR 100%) [4]. ICMR-National Institute of Virology, Pune conducted Nipah surveillance among bat population in North-eastern states of India during 2015. The surveillance revealed the presence of NiV among P. medius bats collected from Cooch Bihar district, West Bengal and Dhubri district, Assam [5]. Similarly, Nipah surveillance was carried out amongst pig population of eight districts of Mizoram state in North-East. However, all the pig serum samples tested negative for anti-Nipah IgG antibodies [6]. The recent emergence of NiV was reported from Kerala state in the Western Ghats during 2018–2019 [7], [8]. Investigation of 2018 outbreak revealed the exposure of the index case to the bats and subsequent human to human transmission [9]. Further, the source of infection in a single case of NIV infection in Kerala during 2019 was also linked to bats. However, the exact cause of the spillover event remained untraced. Studies in bat populations at the locations associated with the index case confirmed presence of NiV RNA and anti-IgG NiV antibodies [10]. The sequence analyses showed its deviation from the NiV strains from Bangladesh and North-eastern region of India. Kerala is the southernmost state of India, distinctly located far away from the earlier NiV hot-spots. Knowledge of the distribution and movement patterns of bat species that act as the reservoir hosts of Nipah virus is necessary to identify the regions at risk, possible events of spillover and the role of eco-geographical factors implied in NiV dynamics in bat population. Considering all these factors, the present study was carried out to determine presence of NiV activity in bat populations in southern states and union territories which are geographically close to the new hotspot of NiV and a state (Odisha) bordering West Bengal in the south east. 2 Materials and methods An approval to carry out the present study was obtained from the Institutional Biosafety Committee and Institutional Animal Ethics Committee of ICMR-NIV, Pune, India. Prior to initiating the work, permission was also obtained for trapping the bats from Principal Chief Conservators of Forest of the respective states/Union Territories. ICMR-NIV, Pune team visited the pre-identified areas of different states/UTs to locate the bat colonies. P. medius and R. leschenaultii bats were captured from pre-identified sites in five states (Telangana, Karnataka, Tamil Nadu Kerala and Odisha) and one union territory (Puducherry) respectively using the methodology described earlier [9]. After species identification of trapped bats, throat and rectal swab specimens were collected in virus transport medium (HiViral™ Transport Medium, HIMEDIA) and stored immediately in dry ice. Blood samples (2–3 ml) were collected from the wing (cephalic) vein and serum was separated. After recovery from anesthesia, the bats were released back. Utmost precautions were taken while handling bats and during specimen collections as per defined biosafety protocols and safety practices developed by ICMR-NIV, Pune. A total of 573 throat swabs/rectal swabs of P. medius (n = 541) and R. leschenaultii bats (n = 32) and blood samples of P. medius bats (n = 255) were collected from Telangana, Kerala, Karnataka, Tamil Nadu, Odisha and Puducherry during January-November 2019 [ Table 1]. Throat/rectal swab specimens were screened using Nipah specific real-time reverse transcription–polymerase chain reaction (rRT-PCR) as per the procedure described elsewhere [5]. Serum specimens were heat inactivated at 56°C for 30 min and further tested for the presence of anti-NiV IgG antibodies by indigenously developed Enzyme-Linked Immunosorbent assay (ELISA) as described earlier [10].Table 1 Details of bat capturing sites and anti-Nipah IgG positivity in different states and union territory in India. Table 1Sr. No. Month and year of bat capture District/State Site Species of bats collected (n) NiV Positive/Total TS/RS tested by Real Time RT-PCR NiV IgG Positive /tested by Anti-Nipah IgG ELISA 1 August 2019 Karnataka Directorate of Health and Family Welfare Services, Anandrao Circle, Gandhi Nagar 580008, Bengaluru GPS 12.981179,77574204 P. medius (n = 6) 0/6 NA Site 1- Anamanahalli P. medius (n = 56) 0/56 20/56 Site 2- BPL Industry P. medius (n = 13) 0/13 2/13 Site 3- Koornagere State Forest P. medius (n = 1) 0/1 0/1 Site 4 - Anche, Chittana Halli P.medius (n = 6) 0/6 2/6 Total P.medius (n = 82) 0/82 24/76 2 April 2019 Kozhikode/Kerala Pallikkunnu, Panthirikkara, Near Changaroth LP& UP School, Perambra Town, Kozhikode (April 2019) P. medius (n = 73); R. leschenaultii (n = 10) 0/83 NA Kuttiady (Opposite to Kuttiady Bar, Adjacent to Citizens’ Club, Kuttiady Town), Kozhikode (April 2019) P. medius (n = 1) 0/1 NA November 2019 Site 1 (Perambra) MUP School Road, Changaroth, Perambra, Kozhikode P. medius (n = 25) 0/25 4/24 Site 2 Kuttiady, Maruthonkera bridge, Kozhikode P. medius (n = 2) 0/2 1/2 Site 3 Olipram Kadavu road, Kozhikode P. medius (n = 1) 0/1 1/1 November 2019 Malappuram, Kerala Site 1 (Karakunnu Vandoor) P. medius (n = 1) 0/1 0/1 Site 2 P. medius (n = 7) 0/7 0/7 Idukki, Kerala [during June 2019 outbreak] Site 1 (Near River Thodupuzha, Dist. Idukki, Kerala) P. medius (n = 54) 0/54 4/23 Site 2 (Muttom, Thodupuzha, Dist. Idukki, Kerala) P. medius (n = 8) 0/8 0/8 Site 3 (Aluva, Dist. Idukki, Kerala) P. medius (n = 61) 0/61 6/39 Site 4 (Thuruthipuram, Thekekkara Panchayat) P. medius (n = 4) 0/4 1/4 Site 5 (Vavakkad, Paravoor, Thekekkara Panchayat) P. medius (n = 14) 0/14 1/4 Total P. medius (n = 251); R. leschenaultii (n = 10) 0/261 18/113 3 July and August 2019 Tamil Nadu Site 1: Near Shivmandir Thisupudale marudur (Western Ghat) P.medius (n = 55) 0/55 07/26 Site 2: Courtallam, Parasakthi Women College Tirunevelli, Tamil Nadu P. medius (n = 4) 0/4 NA Site 3: SN College, Madhurai, Tamil Nadu P. medius (n = 7) 0/7 NA Total 0/66 7/26 4 July 2019 Puducherry Site 1: GPS 11.929138.79.8177744412 Urban Forest P. medius (n = 23) 0/23 2/21 Total P. medius (n = 23) 0/23 2/21 5 February and March 19 Sangareddy/ Telangana Manjira Wildlife Sanctuary, Sangareddy P. medius (n = 8) 0/8 NA Rangareddy/ Telangana Premises near Chilkur Balaji Temple, Chilkur, Moinabad Mandal, Rangareddy P. medius (n = 81); R. leschenaultii (n = 4) 0/85 NA Jangaon/ Telangana Premises of the Senior Civil Judge's court, Jangaon, P. medius (n = 5) 0/5 NA Total P. medius (n = 94); R. leschenaultii (n = 4) 0/98 NA 6 January 2019 Dhenkanal/Odisha Site-1 (Village: Birasagar; Taluka: Kamakhyanagar; District: Dhenkanal; State: Odisa) P. medius (n = 6) 0/6 0/1 Site-2 (Village: Brahmani Devi temple, Badamgiri, Kushida; Distirict: Dhenkanal; State: Odisa) P. medius (n = 35) 0/35 0/15 R. leschenaultii (n = 2) 0/2 NA Total P. medius (n = 41); R. leschenaultii (n = 2) 0/43 NA *TS-Throat swab, RS-Rectal swab, n-Number, NA-Not available 3 Results All the throat swabs/rectal swab specimens of P. medius and R. leschenaultii bats were found negative for the NiV RNA using NiV specific rRT-PCR. However, anti-NiV IgG antibodies were detected in serum samples of Pteropus bats collected from Karnataka (24/76), Kerala (18/113), Tamil Nadu (7/26) and Puducherry (2/21). High antibody prevalence of 20% was detected in Pteropus bat species. Anti-NiV IgG antibodies couldn’t be detected in Pteropus serum samples collected from two sites in Dhenkanal, Odisha. Sites of sample collection and anti-NiV IgG positivity in different states and a union territory are depicted in Table 1. 4 Discussion An earlier report indicated anti-NiV IgG positivity in recaptured fruit bats related to Pteropus spp. in Madagascar [11]. The reintroduction of infected bats or the waning or loss of antibodies over time in bats can result in variations in antibody level in a bat population roosting in the same geographical unit. This phenomenon supports our findings of presence of anti-NiV IgG in only a few bat specimens sampled from the colonies in the same geographical area. Similar observations were also reported in Bangladesh, where the presence of anti-NiV IgG with varied seroprevalance was observed in various locations during the six years of study. Besides this, NiV RNA could be detected in a very few bat samples [12], [13]. Migratory behavior is comparatively less common in bats of tropical and sub tropical region as compared to the temperate region and is never associated with hibernation. Migration is mainly driven by the food resources [13]. The paucity of information on bat migration patterns in India limits the knowledge on the migratory bat populations. Phylogenetic analysis of NiV N gene sequences from the new hot spots in Kerala showed grouping into a cluster distinct from the sequences from Malaysia and Bangladesh suggesting the presence of a new genotype independently evolving in Southern India [9]. Besides this, a fatal case of Nipah was recently reported from Kozikode, Kerala during September 2021 for the second time over a period of three years. The evolving epidemiology of NiV infections, varied involvement of other animal species as intermediate hosts and source of infection in human beings in different geographical regions warrants detailed nationwide proactive surveillance for Nipah with “One Health” approach throughout the year in bats and other animals. This would be of immense importance in understanding the dynamics of NiV in the geographical region and might inform about effective strategies for the prevention and control of Nipah. 5 Conclusion The fruit bats of genus Pteropus are identified to be the main reservoir of the nipah virus causing annual outbreaks in Malaysia, Bangladesh and other countries in South-East Asia including India. Three incidences of NiV infections in humans in Southern state of Kerala which is far distant from the known “Nipah belt” in consecutive years with no identified intermediate animal host or confirmed mode of entry into human population warrants the heightened need of constant surveillance of NiV in bats, animals and humans. Author Contributions PDY, DTM, MG, AS contributed to study design, data analysis, interpretation and writing and critical review. BM, RB, UPT, SM, DRP, AMS, SD, GC, SG, MH, ATS, PS, RJ, DYP contributed to data collection and interpretation. PDY, MG, AS, AMS, DYP contributed to the critical review and finalization of the paper. Financial support & sponsorship Financial support was provided by funding of 10.13039/501100001411 Indian Council of Medical Research (ICMR), New Delhi, India at ICMR-National Institute of Virology, Pune under project ‘Countrywide survey of Nipah virus in Pteropus bats’. Acknowledgement Authors gratefully acknowledge the encouragement and support extended by Prof. (Dr.) Balram Bhargava, Secretary, Department of Health Research, Ministry of Health & Family Welfare & Director-General, ICMR, New Delhi and Dr. Priya Abraham, Director, ICMR-NIV, Pune for their constant support. We thankfully acknowledge Mrs. Savita Patil, Mrs. Triparna Majumdar, Ms. Pranita Gawande, Ms. Kaumudi Kalele, Mr. Deepak Mali and Mr. Abhimanyu Kumar, staff of ICMR-NIV, Pune for their excellent technical support. We express our sincere gratitude to the Principal Conservators of Forests of Karnataka, Kerala, Tamil Nadu, Telangana, Puducherry and Odisha and other concerned officials for their permission to capture bats and support during this study. Conflicts of Interest The authors do not have any conflict of interest. ==== Refs References 1 Chua K.B. Goh K.J. Wong K.T. Kamarulzaman A. Tan P.S.K. Ksiazek T.G. Zaki S.R. Paul G. Lam S.K. Tan C.T. Fatal encephalitis due to Nipah virus among pig-farmers in Malaysia Lancet 354 9186 1999 1257 1259 10520635 2 Gurley E.S. Spiropoulou C.F. de Wit E. Twenty years of Nipah virus research: where do we go from here? J. Infect. Dis. 221 Supplement_4 2020 S359 S362 32392321 3 Chadha M.S. Comer J.A. Lowe L. Rota P.A. Rollin P.E. Bellini W.J. Ksiazek T.G. Mishra A.C. Nipah virus-associated encephalitis outbreak, Siliguri, India Emerg. Infect. Dis. 12 2 2006 235 16494748 4 Arankalle V.A. Bandyopadhyay B.T. Ramdasi A.Y. Jadi R. Patil D.R. Rahman M. Majumdar M. Banerjee P.S. Hati A.K. Goswami R.P. Neogi D.K. Genomic characterization of nipah virus, west bengal, India Emerg. Infect. Dis. 17 5 2011 907 21529409 5 Yadav P. Sudeep A. Gokhale M. Pawar S. Shete A. Patil D. Kumar V. Lakra R. Sarkale P. Nichol S. Mourya D. Circulation of Nipah virus in Pteropus giganteus bats in northeast region of India, 2015 Indian J. Med. Res. 147 3 2018 318 29923524 6 Mourya D. Yadav P. Rout M. Pattnaik B. Shete A. Patil D. Absence of Nipah virus antibodies in pigs in Mizoram State, North East India Indian J. Med. Res. 149 5 2019 677 31417037 7 Arunkumar G. Chandni R. Mourya D.T. Singh S.K. Sadanandan R. Sudan P. Bhargava B. Outbreak investigation of Nipah virus disease in Kerala, India, 2018 J. Infect. Dis. 219 12 2019 1867 1878 30364984 8 Sahay R.R. Yadav P.D. Gupta N. Shete A.M. Radhakrishnan C. Mohan G. Menon N. Bhatnagar T. Suma K. Kadam A.V. Ullas P.T. Experiential learnings from the Nipah virus outbreaks in Kerala towards containment of infectious public health emergencies in India Epidemiol. Infect. 2020 148 9 Paul L. Nipah virus in Kerala: a deadly Zoonosis Clin. Microbiol. Infect. 24 10 2018 1113 1114 29935330 10 Sudeep A.B. Yadav P.D. Gokhale M.D. Balasubramanian R. Gupta N. Shete A. Jain R. Patil S. Sahay R.R. Nyayanit D.A. Gopale S. Mourya D.T. Detection of Nipah virus in Pteropus medius in 2019 outbreak from Ernakulam district, Kerala, India BMC Infect. Dis. 21 1 2021 1 7 33390160 11 Brook C.E. Ranaivoson H.C. Broder C.C. Cunningham A.A. Héraud J.M. Peel A.J. Gibson L. Wood J.L. Metcalf C.J. Dobson A.P. Disentangling serology to elucidate henipa‐and filovirus transmission in Madagascar fruit bats J. Anim. Ecol. 88 7 2019 1001 1016 30908623 12 Epstein J.H. Anthony S.J. Islam A. Kilpatrick A.M. Khan S.A. Balkey M.D. Ross N. Smith I. Zambrana-Torrelio C. Tao Y. Islam A. Nipah virus dynamics in bats and implications for spillover to humans Proc. Natl. Acad. Sci. USA 117 46 2020 29190 29201 33139552 13 T.H. Fleming, Bat migration, in: Encyclopedia of Animal Behavior, 2019, 605.

35390635

PMC9754148

NO-CC CODE

2022-12-16 23:26:27

no

Comp Immunol Microbiol Infect Dis. 2022 Jun 26; 85:101800

utf-8

Comp Immunol Microbiol Infect Dis

10.1016/j.cimid.2022.101800

oa_other

==== Front Int Rev Law Econ Int Rev Law Econ International Review of Law and Economics 0144-8188 1873-6394 Elsevier Inc. S0144-8188(21)00013-2 10.1016/j.irle.2021.105989 105989 Article How has the Covid19 pandemic impacted the courts of law? Evidence from Brazil Castelliano Caio a Grajzl Peter bc⁎ Watanabe Eduardo d a University of Brasilia, Brazil b Washington and Lee University, USA c CESifo, Germany d University of Brasilia, Brazil ⁎ Corresponding author at: Department of Economics, The Williams School of Commerce, Economics and Politics, Washington and Lee University, 204 West Washington St., Lexington, VA, 24450, USA. 31 3 2021 6 2021 31 3 2021 66 105989105989 1 3 2021 28 3 2021 29 3 2021 © 2021 Elsevier Inc. All rights reserved. 2021 Elsevier Inc. Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. We provide empirical insight into the consequences of the Covid19 pandemic for the administration of justice. Drawing on a comprehensive monthly panel of Brazilian labor courts and using a difference-in-difference approach, we show that the pandemic has had a large and persistent deleterious effect on adjudicatory efficacy, leading to a massive decrease in the clearance rate and an increase in court backlogs. The pandemic has affected how courts dispose adjudication cases, expectedly causing a plummeting in the share of disputes resolved via trial hearings and, less predictably, exerting a temporally non-linear effect on the share of in-court settlements. Notably, we find no evidence of an effect of the pandemic on efficacy in enforcement. Although the pandemic led to an increase in the share of new filings requiring enforcement, any effect on the relative use of enforcement to execute court-ordered payments has been intermittent and temporary. The intensity of the pandemic has been an important moderating factor. Keywords Covid19 Courts Brazil Labor justice Adjudication Enforcement ==== Body pmc1 Introduction The Covid19 pandemic has fundamentally impacted virtually every facet of our existence, wreaking havoc in healthcare systems, leading to a massive death toll, and causing a profound economic and social disruption. Unsurprisingly, the functioning of judicial systems has been impacted as well. Across jurisdictions worldwide, lockdowns, self-isolation, and restrictions on the population's movement and assembly have affected the ability of the courts to perform their function. Summarizing the state of affairs at the onset of the pandemic, the European Commission for the Efficiency of Justice, for example, noted that “the courts are facing numerous challenges to remain operational due to lack of personnel, hearings are being cancelled, and access to justice is temporarily limited” (CEPEJ, 2020a). Yet despite the agreement among policymakers, court administrators and scholars about the pressing nature of the problem1 , we are aware of no systematic quantitative study characterizing the impact of the pandemic on the output and performance of the courts of law. In this paper, we provide the first full-fledged empirical inquiry into the effect of the Covid19 pandemic on the success of courts at performing their primary function, the resolution of disputes and disposition of cases. To this end, we draw on comprehensive data on the activity of labor courts in Brazil. In Brazil, the impact of the pandemic has been especially severe.2 At the same time, the court system as a whole, and labor courts in particular, have been subject to a series of pandemic-instigated policy responses that have fundamentally affected their operations. Brazilian labor courts possess jurisdiction over all labor law-related disputes, with the overarching majority of filed claims pertaining to infringements of employee-employer agreements, especially with regard to unlawful termination of employment and unpaid compensation. As such, labor courts constitute a crucial component of the Brazilian judicial system. Moreover, empirical insight into the effect of the pandemic on the administration of Brazilian labor justice provides valuable clues about the effects of the pandemic on the courts of law in other jurisdictions worldwide. Our data are unique in several respects. First, the data cover all Brazilian regional labor courts observed at a monthly level, a time frequency that is higher than that normally encountered in the empirical literature on court administration. Second, we observe the regional labor courts both in the era prior to and after the onset of the pandemic, a feature of the data that aids the assessment of the impact of the pandemic. Third, our dataset encompasses the indicators of court activity with regard to both adjudication and enforcement. Unlike the resolution of adjudication cases, the disposition of enforcement cases does not entail the conduct of hearings. Nevertheless, the enforcement of final decisions (res judicatae) requires judicial effort, represents a very significant share of dockets of the Brazilian labor courts, and, when required, constitutes a critical step in the process of restoration of justice for the prevailing party. Utilizing the resulting panel dataset, in which a unit of observation is a regional labor court in a given month, we are therefore able to offer an in-depth empirical glimpse into how the pandemic has affected the administration of justice in one prominent legal system. The assessment of the consequences of the pandemic, however, presents an empirical challenge. A naïve pre- versus post-pandemic comparison of court outcomes, for example, will not adequately capture the effect of the pandemic when court activity is subject to inherent temporal trends. We therefore adopt a difference-in-difference (DD) approach. To construct the missing counterfactuals depicting the changes in the labor court outcomes in the absence of the pandemic, we rely on the changes in the outcomes of the same set of courts observed during the pre-Covid19 epoch. That is, prior to March―the month of the onset of the pandemic in the year 2020―the trends in the outcomes in the Brazilian regional labor courts during the Covid19 epoch, defined as the 12-month span from November 2019 to October 2020 (the latest month of data availability), closely resemble those registered one year earlier. This feature renders the same set of regional labor courts that we observe also during the pre-Covid19 epoch, defined as the 12-month span from November 2018 to October 2019, a suitable control group for purposes of our analysis. Our DD estimates show that the Covid19 pandemic, above all, drastically adversely affected the adjudicative efficacy of Brazilian labor courts. We find that the pandemic-caused average drop in the monthly clearance rate in adjudication is of the size of about one-fifth of the mean monthly clearance rate during the pre-pandemic era. Consequently, the pandemic led to a surge in court backlogs in adjudication, with the estimated increase in the number of pending cases per judge equal to about a quarter of the mean monthly pending cases per judge recorded during the pre-pandemic era. Thus, the pandemic is expected to exert a lasting deleterious effect on the ability of the courts to deliver justice in due time. At the same time, the pandemic has impacted the modes with which the courts dispose adjudication cases. As one would have expected, the share of adjudication cases disposed via trial hearings decreased sharply as the courts implemented an initial ban on all in-person court activities. The pandemic, however, has at least on average not affected the share of adjudication cases disposed via in-court settlement or lawsuit withdrawal. Interestingly, we also do not find evidence of an impact of the pandemic on court outcomes in enforcement. Neither the clearance rate in enforcement nor the volume of pending enforcement cases per judge exhibit a detectable change as a result of the pandemic. Similarly, even though the pandemic resulted in an increase in the relative demand for judicial enforcement versus adjudication, we do not find evidence that the pandemic altered the relative reliance on judicial enforcement as means of securing the execution of court-sanctioned payments among the disputing parties. As we emphasize in interpreting our results, this last finding is an especially important and, in many respects, a reassuring one. At least during our observation window, the pandemic has on average not increased the need for judicial enforcement that could have conceivably arisen amidst the pandemic-instigated economic downturn, a period when the losing parties pressed for cash may be tempted to purposely avoid the execution of court-ordered damage and compensation payments. In addition to offering the baseline DD estimates of the average effect of the pandemic, we provide two further sets of estimates. We first explore an event-study approach to provide insight into the month-by-month effect of the pandemic. The response of the Brazilian labor courts to the pandemic has evolved with the evolution of the pandemic itself. Intuitively, one would therefore expect that the effects of the pandemic on court outcomes varied over time. As we demonstrate, the estimates based on a dynamic specification indeed reveal important insights into the temporal effects of the pandemic that would remain hidden if one focused solely on the static DD estimates. Our analysis shows, for example, that the negative effects on the clearance rate in adjudication were largest in the first few months after the start of the pandemic, but gradually diminished in magnitude as time unfolded and the courts implemented various response measures. Similarly, while the static DD estimates do not show an effect of the pandemic on the share of in-court settlements, the estimates based on the dynamic specification uncover that the initial effect on the share of in-court settlements was, in fact, negative, while the effect several months into the pandemic turned to positive. We interpret these findings in light of the existing theories of litigation. Finally, we investigate the heterogeneity of the effect of the pandemic with respect to the intensity of the pandemic. Since its onset, the Covid19 pandemic has in Brazil exhibited considerable geographic and temporal variation with regard to its severity, with different regions adopting different response measures at different points in time. The specificities of local conditions at a given point in time plausibly influenced the functioning of the courts in that locality at that point in time. To measure the intensity of the pandemic, we utilize official Brazilian monthly regional data on new Covid19 infections and Covid19-related deaths. We then estimate a specification where the DD effect of the pandemic is allowed to vary with pandemic intensity. Our findings suggest that the effect of the pandemic on court outcomes exhibits important heterogeneity with respect to pandemic intensity. We find, for instance, that the adverse effect of the pandemic on adjudicatory efficacy has been larger for those labor court region-months that have exhibited stronger pandemic intensity. Similarly, the average effect of the pandemic on the share of in-court settlements is positive in the labor court region-months characterized by a higher incidence of new infections or a greater virus death toll. But we once more do not find evidence of an effect of the pandemic on the core judicial enforcement outcomes. Our paper adds to and links two primary strands of literature. On the one hand, we contribute to the emerging literature on the consequences of the Covid19 pandemic for the courts of law. The Covid19 crisis has stimulated a global policy and academic debate about the numerous ways in which the pandemic has already affected, and will likely still affect, the operations of courts and the delivery of justice (e.g., McIntyre et al., 2020; Baldwin et al., 2020; Puddister and Small, 2020; Warner, 2020; Sourdin and Zeleznikow, 2020; Engstrom, 2020; Pistor, 2020; Matyas et al., 2021). The existing contributions on the topic, however, have been primarily descriptive in character. In particular, the research has not illuminated the impact of the pandemic on the administration of justice using comprehensive court data and rigorous quantitative analysis. To the best of our knowledge, our paper is the first to accomplish this task. At the same time, we contribute to the growing empirical literature on the administration of justice and the functioning of courts. Motivated by the ubiquity of court delays and the corresponding social costs incurred by numerous jurisdictions worldwide, an important subset of this literature has focused on the determinants of the efficacy of courts at performing their core role, the disposition of cases (e.g., Buscaglia and Ulen, 1997; Beenstock and Haitovsky, 2004; Rosales-López, 2008; Dimitrova-Grajzl et al., 2012; Di Vita, 2012; Chemin, 2009; Christensen and Szmer, 2012; Voigt, 2016; Marciano et al., 2019; Bełdowski et al., 2020; Grajzl and Silwal, 2020; Castelliano et al., 2020a). A related subset of research has investigated court-level factors affecting the use of different modes of court case disposition (e.g., Galanter, 2004; Dimitrova-Grajzl et al., 2014). In contrast, judicial enforcement of final verdicts, a distinct and often especially vital facet of court activity, has received much less attention (EBRD, 2014; Castelliano et al., 2020b). Our paper advances the corresponding literature by empirically examining the impact of the Covid19 pandemic on all three of the above-noted aspects of court output and operations. The rest of the paper is organized as follows. Section 2 provides an overview of the Brazilian labor courts and their response to the pandemic. Section 3 introduces our data. Section 4 develops our empirical approach. Sections 5 presents and discusses the results. The final section concludes. 2 Background 2.1 Brazilian labor justice and proceedings Labor justice, administered in the labor courts, is a key pillar of the Brazilian judicial system.3 In 2019, for example, new labor-court case filings represented about eleven percent of all new case filings in Brazilian (state and federal) courts (CNJ, 2020). Labor courts are specialized courts, with labor court judges following a career track that is separate from that followed by other federal and state judges. Brazil is divided into 24 labor court regions. Each labor court region is in turn divided into districts, with each district featuring one or more labor court offices. Each labor court region, however, has a single second-instance labor court, referred to as the regional labor court (tribunal regional do trabalho). In addition to adjudicating appeals to first-instance decisions, the regional labor court exercises administrative authority over the first-instance tribunals in its region. Jurisdictional rules require that labor law-related disputes be adjudicated at a first-instance court office with jurisdiction over the geographic area of the dispute's origination. In districts with more than one labor court office, a newly-filed case is allocated to a particular office using a system of computerized random assignment. Within any labor court office, cases are then allocated between a titled judgeship and a substitute judgeship using an analogous procedure (see Castelliano et al., 2020a). The institutional division of court offices into titled and substitute judgeships is intended to limit the possibilities of the litigants to engage in judicial forum shopping and, at the same time, operationalizes the Brazilian system of judicial career advancement where all new judges are first employed as substitute judges. The overwhelming majority (more than 99 percent) of cases adjudicated in Brazilian labor courts are employee claims stemming from alleged violation of employment contracts in the private sector. Among those, the most commonly brought-up subject issues are the termination of employment and overdue wages. Other frequent subject issues pertain to employer contributions into the public severance indemnity fund, severance payment, overtime wages, premiums for high-risk work, as well as compensation for pain and suffering (TST, 2020). Brazilian labor courts dispose cases in both the adjudication (conhecimento) and the enforcement (execução) stage of court proceedings, an important distinction emphasized by the Brazilian civil procedure and meticulously tracked by official court statistics. Specifically, a case filed at a first-instance labor court that is not settled, withdrawn, or dismissed on procedural grounds is eventually resolved via a court decision. To reach a decision, the adjudicating judge conducts hearings, interviews witnesses, and examines facts. The disputing parties may appeal at the second and higher instance, until the decision attains the status of a final decision. Often, the final decision specifies a monetary transfer from the losing to the winning party. For example, an employer who is found to have unlawfully terminated an employee's employment contract is expected to compensate the employee for the lost wages and any other damages. In those instances, the court orders the losing party to execute the payment to the benefit of the prevailing party. But the losing party sometimes does not comply. In the event of such non-compliance, the winning party must initiate separate judicial enforcement proceedings at the first-instance labor court office that adjudicated the original dispute.4 To enforce court-mandated payments, labor-court judges resort to a variety of patrimonial constrictions, the implementation of all of which requires judicial attention and time (see Castelliano et al., 2020b). Enforcement cases hence constitute a substantial portion of the labor courts' workload and facilitate the transfer of a very significant sum of awarded compensation payments. For example, in year 2019, the aggregate value of payments secured upon the completion of enforcement proceedings was more than three times as large the value of voluntary payments following the final court decisions (TST, 2020). 2.2 Operational response of labor courts to the pandemic In Brazil, as in many other countries, the first instances of Covid19 infections were officially registered in March 2020. Upon the call of the President of the Republic, the Brazilian National Congress formally recognized the state of public calamity on March 20. Even prior to that, on March 18, the National Council of Labor Justice decreed the suspension of all in-person services, and in particular the conduct of trial hearings, effective from March 19. The courts, including the labor courts, were to continue in an uninterrupted fashion only with the provision of essential services (such as information-technology support, facility security, and payroll). Importantly, all procedural deadlines related to case processing (for example, to file an appeal or include new evidence) were suspended until June 14, 2020. From March 19, the judges and their support staff were authorized to carry out remotely all substantive tasks, including the preparation of judgments and the execution of administrative duties. The resulting provision enabled the judges to implement remotely and via virtual sessions all key decision-making activities, with the exception of those that inherently rely on the conduct of trial hearings. The abrupt switch to remote work undoubtedly led to initial disorganization and difficulties. There was, however, also an anticipation that the judges and their staff would sooner or later adapt to the new work format. At the same time the courts continued to accept new filings. In Brazilian labor courts the overwhelming majority of cases had already been filed electronically even prior to the start of the pandemic. For example, in the year 2019, as many as 99 percent of all new labor cases were filed electronically. Therefore, at least when it comes to the logistics of the initiation of new filings, the onset of the pandemic was not expected to radically curtail access to labor justice.5 In addition, an executive order of the President of the Republic, valid from March 22 until July 19, enabled the employers to implement remote work, use individual and collective vacation days, as well defer payments into the public severance indemnity fund with the express aim of avoiding instances of breach of labor contracts. These measures naturally mitigated the demand for labor justice in anticipation of the pandemic-caused economic disruption. Once the inevitability of the prolonged nature of the pandemic became apparent and upon the passage of supporting regulations, on May 5, the National Council of Labor Justice authorized the courts to conduct virtual trial hearings. To take into consideration the local differences in both the intensity of the pandemic and the state and municipal government responses to it, the administrative and technological implementation of the conduct of virtual hearings was delegated to the regional labor courts themselves. The pace of the implementation of virtual hearings was therefore likely not uniform across labor courts and in particular across individual judges. Many judges, but also clients and their attorneys, certainly grappled with the usage of new technology. In addition to enabling the conduct of virtual hearings, the May 5 resolution by the National Council of Labor Justice re-instated the validity of the procedural deadlines that had been suspended since March 19. The judges, however, were granted discretion with regard to the application of the procedural deadlines pertinent to specific cases based on the epidemiological conditions relevant to each case. Finally, the resolution established criminal liability for the execution of any court operations that could potentially contribute to the spread of the virus. That is, the courts themselves were not to, and did not, become a contributor to the spread of the virus. On June 16, the National Council of Labor Justice formally allowed for a gradual reestablishment of in-person court activities, including the conduct of hearings, subject to the adoption of appropriate safety protocols by the courts. Much like in the case of the prior introduction of the possibility to carry out virtual hearings, the implementation of the eventual return to in-person hearings was entrusted to the regional labor courts themselves. Given the raging pandemic, the courts and especially the judges adopted a very cautious approach. Because all Brazilian federal judges, including labor court judges, are civil servants with lifetime tenure, the judges have not felt the pressure to physically appear in their formal office spaces to do their work. At the same time, the vast majority of the judges are proficient in the use of required information technology and managed to adapt to working from home. Thus, in practice, the physical facilities of nearly all courts, including the labor courts, remained closed to the public and the courts have continued to conduct their operations remotely, in a virtual format, even after June. Fig. 1 summarizes the key events pertinent to the functioning of the labor courts from the onset of the pandemic until October 2020, the last month of coverage in our data.Fig. 1 The timeline of the key operational responses of the labor courts to the pandemic. Fig. 1 3 Data The source of our data on regional labor courts is a database compiled by the Brazilian Superior Labor Court. There are 24 labor court regions in Brazil. For purposes of the analysis, we combine two labor court regions, which together encompass the state of São Paulo, into a single labor court region. The combining of these two labor court regions is necessary because consistent monthly Covid19-related data that we employ in a subset of our analysis are available only at the level of the state of São Paulo as a whole.6 For each of the correspondingly-defined 23 regional labor courts, we observe monthly data on court staffing and court activity over a 24-month span between November 2018 and October 2020.7 We split the resulting time span into two contiguous, non-overlapping 12-month subperiods. The first, from November 2019 to October 2020, includes the onset of the Covid19 pandemic in March 2020. We refer to this period as the Covid19 epoch. The second subperiod, from November 2018 and October 2019, covers the exact same months as the Covid19 epoch, but occurring one year earlier when the labor courts were not subject to any noteworthy shocks or legislative changes. We refer to this second subperiod as the pre-Covid19 epoch. As we clarify in Section 4 below, the resulting two-epoch structure of our data facilitates the estimation of the effect of the Covid19 pandemic on labor court outcomes. During each of the two epochs we for each labor court in every month observe outcomes indicative of the extent of court efficacy in the context of both adjudication and enforcement. Specifically, for each of the two types of proceedings, we observe the monthly number of resolved cases and the number of newly filed cases. For purposes of empirical analysis, we divide the number of resolved cases of a given type (adjudication or enforcement) by the number of newly filed cases of the same type. The resulting case type-specific clearance rate is an indicator of the ability of a court to meet the demand for its services in the pertinent domain. As such, clearance rate is a core and commonly utilized measure of court efficacy (see, e.g., CEPEJ, 2020b; Voigt, 2016). A value of the clearance rate greater than one indicates that the court is able to both meet the ongoing demand and reduce existing backlogs. In contrast, a value of the clearance rate smaller than one implies that the court is contributing to the accumulation of case backlogs. For both adjudication and enforcement cases, we also observe the total number of cases still pending at the end of every month at every court. For each court, we divide the total number of pending adjudication and enforcement cases, respectively, with the number of judges serving at the court during the applicable month. The resulting end-of-month number of pending adjudication and enforcement cases per judge, respectively, are direct measures of court backlogs. The combination of the clearance rate and the number of pending cases per judge in adjudication and enforcement, respectively, thus allows us to investigate the impact of the pandemic on court efficacy in adjudication and enforcement. At the adjudication stage, we further observe the monthly number of cases resolved via trial hearings, the number of in-court settlements, and the number of withdrawals. For every court in every month, we divide each of these three variables with the total volume of disposed adjudication cases. The share of cases resolved via trial hearings, the share of case settled in court, and the share of withdrawals are then measures of court output with regard to modes of case disposition. We use thus-defined outcomes to explore the consequences of the pandemic for the courts' modes of disposition of adjudication cases. For every court in each month, we observe the total value of all payments executed as a consequence of court proceedings and the value of payments executed only upon completion of the enforcement proceedings. In general, labor court-sanctioned payments are executed for one of three distinct reasons: a spontaneous transfer that is executed voluntarily by the losing party after a court verdict; an in-court settlement-based or conciliation-induced payment that occurs at any stage of the adjudication or enforcement proceedings; or a payment secured upon the completion of judicial enforcement proceedings if the losing party fails to comply with the final decision. We calculate the monthly share of the value of all executed payments that occur as a consequence of enforcement. The resulting measure is indicative of the relative importance of judicial enforcement for the execution of court-sanctioned payments. Given the decline in the economic activity as a consequence of the pandemic and the corresponding liquidity pressures faced by many employers, we want to examine whether the pandemic has resulted in greater reliance on judicial enforcement. Such a scenario could arise if cash-stripped employers, found to had violated the Brazilian labor law, perhaps purposefully chose to avoid voluntarily executing court-ordered damage and compensation payments. Finally, using the information on the volume of new filings of adjudication and enforcement cases at each court during every month, we compute the share of newly filed enforcement cases in all (adjudication and enforcement combined) newly filed cases. The resultant variable, a measure of the composition of new filings, allows us to gauge whether, and if so to what extent, the pandemic and the ensuing economic crisis have altered the balance between the demand for enforcement versus demand for adjudication. The data, however, do not allow us to observe the temporal evolution of the exact composition of newly-filed and processed adjudication and enforcement cases, respectively. Consequently, we are unable to ascertain, for example, to what extent any pandemic-induced changes in labor-court efficacy in adjudication and enforcement, respectively, are purely due to the response of the courts per se versus any changes in the complexity of the underlying cases or, when it comes to the clearance rate, perhaps even the strategic behavior of the litigants. Thus, while operational responses of the courts alone have undoubtedly been of central importance, our results should be interpreted as amalgamating multiple mechanisms. We merge the data for the 23 above-defined labor courts with the official monthly data on the incidence of new Covid19 cases and Covid19-related deaths, recorded at the geographic level of labor court regions and normalized using Brazilian census population data.8 This gives us monthly per-capita measures of the intensity of the Covid19 pandemic at the level of each labor court region. From March 2020 onwards, the resulting variables vary both from month to month and across the labor court regions. Table 1 shows the basic descriptive statistics for the outcome and selected additional variables that we use in the analysis for both the Covid19 epoch (part A) and the pre-Covid19 epoch (part B). Table A1 in the Appendix provides the corresponding variable definitions.Table 1 Descriptive statistics. Table 1 Part 1: Covid19 epoch (Nov 2019-Oct 2020) Part A1: Nov 2019-Feb 2020 Part A2: Mar 2020-Oct 2020 Obs. Mean S.D. Obs. Mean S.D. Courts, adjudication  Clearance rate (adj.) 92 1.0871 0.2145 184 0.8662 0.2553  Pending per judge (adj.) 92 235.3 94.3 184 260.7 102.3  Share resolved in hearings 92 0.2952 0.0877 184 0.1808 0.1003  Share settled 92 0.3511 0.0754 184 0.3659 0.1240  Share withdrawn 92 0.0428 0.0305 184 0.0418 0.0201 Courts, enforcement  Clearance rate (enf.) 92 1.2727 0.6163 184 1.3766 0.5462  Pending per judge (enf.) 92 611.2 214.0 184 599.3 239.8  Share enforced payments 92 0.4377 0.1338 184 0.4689 0.1551 Courts, staffing  Staff per judge 92 5.12 1.05 184 5.06 1.03 Courts, new filings composition  Share enforcement new filings 92 0.3147 0.0740 184 0.3831 0.0921 Pandemic intensity  New Covid19 cases per 1000 people 92 0 0 184 3.84 3.41  Covid19 deaths per 10,000 people 92 0 0 184 0.97 0.84 Part B: Pre-Covid19 epoch (Nov 2018-Oct 2019) Part B1: Nov 2018-Feb 2019 Part B2: Mar 2019-Oct 2019 Obs. Mean S.D. Obs. Mean S.D. Courts, adjudication  Clearance rate (adj.) 92 1.1829 0.2142 184 1.2120 0.1683  Pending per judge (adj.) 92 306.6 139.8 184 260.3 118.1  Share resolved in hearings 92 0.2815 0.1251 184 0.3305 0.0831  Share settled 92 0.3422 0.0756 184 0.3687 0.0703  Share withdrawn 92 0.0385 0.0216 184 0.0410 0.0256 Courts, enforcement  Clearance rate (enf.) 92 0.9500 0.3241 184 0.9846 0.4896  Pending per judge (enf.) 92 608.4 217.1 184 602.7 221.6  Share enforced payments 92 0.4613 0.1387 184 0.4607 0.1295 Courts, new filings compositon  Share enforcement new filings 92 0.3216 0.0560 184 0.3291 0.0532  Courts, staffing  Staff per judge 92 5.26 1.12 184 5.11 1.04 Notes: The table presents descriptive statistics for the eight outcome variables used in the empirical analysis, a control (staff per judge), and moderating variables (on pandemic intensity). Observation is a labor court (or, equivalently, labor court region) in a given month. 4 Empirical approach To obtain an initial glimpse into the consequences of the pandemic for court outcomes, one could imagine pursuing two simple approaches. Under one approach, one might contrast the outcomes post March with the outcomes prior to March during the Covid19 epoch, that is, compare the mean for the outcomes of interest in part A2 of Table 1 with the mean for the same outcomes in part A1 of Table 1. The resulting approach, however, does not take into account the trends in the data. For instance, the pre-March period subsumes the holiday-season months of December and January, when court activity naturally slows down every year. The post-March versus pre-March comparison alone would therefore unlikely yield a compelling estimate of the effect of the pandemic. Alternatively, one might contrast the post-March outcomes from the Covid19 epoch with the post-March outcomes from the pre-Covid19 epoch, that is, compare the mean for the outcome in part A2 of Table 1 with the mean for the same outcomes in part B2 of Table 1. Yet the comparison of the post-March outcomes from the Covid19 epoch with the post-March outcomes from the pre-Covid19 epoch does not address the concern that the Covid19 and pre-Covid19 epochs plausibly differ in unobserved ways, which confounds the estimate of the effect of the pandemic. To address the deficiencies inherent in the two naïve approaches described above while at the same time combining their intuitively-appealing features, we use a difference-in-difference approach and exploit the exogenous nature of the pandemic. We first posit the following general model:(1) ycet=α+βPostt×Covide+γCovide+λt+μce+δSPJcet+εcet, where c denotes labor court, e epoch (pre-Covid19 or Covid19), and t month (from November to October the following year). ycet is one of the eight outcome variables, listed in the first eight entries in Table 1 or Table A1 in the Appendix. Postt is a dummy equal to 1 if the observation is from the month of March or later. Covide is a dummy equal to 1 if the observation is from the Covid19 epoch and 0 if it is from the pre-Covid19 epoch. The month fixed effect λt fully absorbs the timing of the observation relative to the start of the Covid19 pandemic, rendering a separate inclusion of Postt on the right-hand side of (1) redundant. μce is the labor court-in-epoch fixed effect, which absorbs the time-invariant average impact on the outcome under consideration of each labor court during each epoch. SPJcet is the staff per judge control that varies across labor courts, epoch, and months. εcet is the error term. The coefficient of interest in expression (1) is β, the difference-in-difference (DD) estimate of the impact of the Covid19 pandemic―our treatment of interest―on the court outcome under consideration. More precisely, β captures the difference between the post-March versus pre-March change in the court outcome under consideration during the Covid19 epoch and the analogous change during the pre-Covid19 epoch. That is, the first difference contrasts the post-March court outcome with the pre-March outcome in the Covid19 epoch, our treated group. However, we do not observe how the court outcome under consideration would have looked like after March 2020 had the pandemic never occurred. To construct the pertinent counterfactual, we use the change in the post-March versus pre-March outcome during the pre-Covid19 epoch, our control group. Subtracting this second difference from the first difference then provides a difference-in-difference estimate of the effect of the pandemic on the court outcome under scrutiny. In order for the labor courts in the pre-Covid19 epoch to serve as a good control group for the labor courts in the Covid19 epoch, in the absence of the pandemic the change in the court outcomes would have to be the same for both groups. Fig. 2 indicates that most court outcomes under consideration prior to March during the Covid19 and pre-Covid19 epochs indeed exhibit substantially parallel comovements, an indication that the parallel trends assumption would seem an apposite one to make in our context. In Section 5.2, we provide additional evidence in favor of the parallel trends assumption by using a dynamic specification and demonstrating the lack of pre-trends for the vast majority of the featured outcomes. Importantly, we have also verified that none of our findings are sensitive to an alternative definition of the pre-Covid 19 epoch.9 Nevertheless, the reader should keep in mind that the parallel trends assumption is inherently untestable. Our empirical strategy would be flawed if, for example, the labor courts during the pre-Covid19 epoch experienced unsuspected post-March shocks. In response to such concerns we emphasize that, based on our careful scrutiny of the Brazilian labor justice system during the time period under consideration, we did not discover any noteworthy events or developments that could invalidate our DD approach. In particular, from the start of the chosen pre-Covid19 epoch to the onset of the pandemic in 2020, the labor courts were not subject to any major institutional changes. Institutional history of the Brazilian labor justice system therefore lends credibility to our proposed methodology.Fig. 2 Temporal trends in court outcomes by epoch. Notes: The figure depicts the temporal evolution of the cross-sectional means, computed at the level of a regional labor court, of depicted variables for the Covid19 epoch (Nov 2019-Oct 2020) and the pre-Covid19 epoch (Nov 2018-Oct 2019). Fig. 2 The estimate of β based on specification (1) is informative of the average effect of the pandemic on the considered court outcomes. To gain further insight into whether, and if so how, the effect of the pandemic has varied over time, we estimate the following specification:(2) ycet=α+∑τβτMonthτ×Covide+γCovide+λt+μce+δSPJcet+εcet, where Monthτ is a dummy equal to 1 if the observation is from a specific month τ from the set of twelve months from November to the following October. As the month with respect to which we compare all month-by-month effects, we omit the month of February, that is, the month immediately preceding the start of the pandemic affecting the Covid19 epoch. The remaining elements of the right-hand-side of expression (2) are as defined under expression (1). Finally, we leverage the fact that, while the pandemic hit Brazil hard, not all regions were affected to the same extent and at the same time. We are therefore able to explore whether the effect of the pandemic on court outcomes has varied with the intensity of the pandemic as captured by the incidence of new Covid19 infections or Covid19-related deaths. To this end, we also examine the results based on the following model:(3) ycet=α+θPostt×Covide×Incidencecet + βPostt×Covide+γCovide+λt+μce+δSPJcet+εcet. In expression (3), Incidencecet is either the per-capita number of new Covid19 infections or the per-capita number of Covid19-related deaths in labor court region c during epoch e in month t. The remaining components of the right-hand-side of expression (3) coincide with those in expression (1). The primary coefficient of interest is θ, capturing the extent to which the effect of the pandemic on court outcome under investigation has varied with the intensity of the health crisis. We estimate all models using OLS. We base inference on heteroscedasticity-robust standard errors clustered at the level of the regional labor court. We therefore allow for correlation of unobservables within each regional labor court over time, an important feature given that we observe the same set of regional labor courts during both the Covid19 and the pre-Covid19 epoch. Because of a relatively small number of clusters (23), we for each coefficient estimate of interest also report the p-value for the test of the null of no effect, executed using wild bootstrap with 1000 replications (see Roodman et al., 2019). 5 Results and discussion 5.1 Average effect The results on the average effect of the pandemic are reported in Table 2 through 5. In each table, the odd-numbered columns report the results for the specification without any non-essential fixed effects or controls. The even-numbered columns show the results for the specification that includes court-in-epoch fixed effects and the staff per judge control. As anticipated given the exogenous nature of the pandemic, for each outcome, both specifications yield congruent estimates. In interpreting the results, we focus on the specifications featuring the full set of court-in-epoch fixed effects and the staff per judge control.Table 2 Effect on efficacy in adjudication. Table 2 Outcome: Clearance rate (adj.) Outcome: Pending per judge (adj.) (1) (2) (3) (4) Post March × Covid19 epoch −0.2243*** −0.2171*** 66.4289*** 62.3226*** (0.0432) (0.0457) (9.1613) (8.9006) [<0.000] [<0.000] [<0.000] [<0.000] Covid19 epoch FE Yes Yes Yes Yes Month FE Yes Yes Yes Yes Court-in-epoch FE No Yes No Yes Staff per judge control No Yes No Yes Observations 552 552 552 552 R-squared 0.3888 0.5779 0.0285 0.9503 Notes: The table presents OLS results based on the estimation of model (1). Post March is a dummy equal to one if observation is from the month of March or later in the applicable epoch. Covid19 epoch is a dummy equal to one if observation is from the Nov 2019-Oct 2020 period. Standard errors in parentheses are clustered at the level of regional labor court. In the brackets are pvalues for the test of the null of no effect executed using wild bootstrap with 1000 replications ***, **, and * denote statistical significance at the 0.1 %, 1%, and 5% level, respectively. Table 2 presents the results on the average effect of the pandemic on court efficacy in adjudication. The chief finding is apparent: the pandemic resulted in a quantitatively very large decrease in court efficacy in adjudication. Based on our estimates, the clearance rate in adjudication on average decreased by 0.22, an effect equal to about 20 percent of the average monthly clearance rate attained in the months from November to February in the Covid19 epoch or 18 percent of the average monthly clearance rate attained between March and October in the pre-Covid19 epoch. Notably, part (b) of Fig. 2 shows that the onset of the pandemic brought a decrease in the number of newly filed adjudication cases. The drop in the clearance rate in adjudication, implied by our estimates, is therefore not a consequence of an increase in demand for court services. Rather, the decrease in the courts' ability to meet the demand for adjudicatory services is driven by a fall in judicial output, as indicated in part (a) of Fig. 2. The large fall in the clearance rate in adjudication after March of the Covid19 epoch resulted in a post-March average monthly clearance rate notably smaller than one (see part A2 of Table 1), the critical value at which a court is just able to meet the incoming demand. Concurrently with the plummeting of the clearance rate, the number of pending adjudication cases per judge thus on average increased by as much as 62, an effect equal to about 26 percent of the average number of pending adjudication cases per judge in the months from November to February in the Covid19 epoch or 24 percent of the average number of pending adjudication cases per judge between March and October in the pre-Covid19 epoch. This is evidence that the pandemic resulted in a substantial increase in court backlogs. Table 3 shows the results on the average effect of the pandemic on the mode of disposition of adjudication cases. As one would have expected, we find a very sizeable effect of the pandemic on the share of cases resolved in trial hearings. Based on our estimates, the share of cases resolved in trial hearings on average decreased by 0.14, an effect equal to about 47 percent of the average monthly share of cases resolved in trial hearings in the months from November to February in the Covid19 epoch or 42 percent of the average monthly share of cases resolved in trial hearings between March and October in the pre-Covid19 epoch. We, however, do not find evidence that the pandemic, at least on average, affected the share of in-court settlements or the share of withdrawals, respectively.Table 3 Effect on modes of case disposition in adjudication. Table 3 Outcome: Share resolved in hearings Outcome: Share settled in-court Outcome: Share withdrawn (1) (2) (3) (4) (5) (6) Post March × Covid19 epoch −0.1365*** −0.1360*** −0.0164 −0.0169 −0.0037 −0.0042 (0.0270) (0.0282) (0.0098) (0.0107) (0.0044) (0.0046) [<0.000] [<0.000] [0.101] [0.108] [0.480] [0.425] Covid19 epoch FE Yes Yes Yes Yes Yes Yes Month FE Yes Yes Yes Yes Yes Yes Court-in-epoch FE No Yes No Yes No Yes Staff per judge control No Yes No Yes No Yes Observations 552 552 552 552 552 552 R-squared 0.3975 0.5980 0.2123 0.7132 0.0242 0.7832 Notes: See notes under Table 2. Table 4 displays the results on the average effect of the pandemic on judicial enforcement outcomes. In contrast to the profound detrimental effect of the pandemic on court efficacy in adjudication, we find no effect of the pandemic on court efficacy in enforcement. Neither the effect on the clearance rate in enforcement nor the effect on the volume of pending enforcement cases per judge are statistically significantly different from zero. These findings may be explained by three key features of the Brazilian labor-court enforcement proceedings. First, enforcement proceedings do not require the conduct of hearings, rendering enforcement comparatively less vulnerable to the pandemic-caused disruption. Second, labor-court judges are able to implement many enforcement-related tasks, such as the confiscation of funds from the checking account of a non-compliant losing party or the blocking of the transfer of their property title, using remotely-accessible software (see Castelliano et al., 2020b). Third, it is likely that, following the onset of the pandemic, the judicial and staff effort that would have normally been devoted to the conduct of trial hearings and other adjudication activities was purposefully re-directed toward enforcement activities.Table 4 Effect on judicial enforcement. Table 4 Outcome: Clearance rate (enf.) Outcome: Pending per judge (enf.) Outcome: Share enforced payments (1) (2) (3) (4) (5) (6) Post March × Covid19 epoch 0.0495 0.0573 −3.4881 −11.1254 0.0256 0.0229 (0.1336) (0.1385) (11.8367) (10.9191) (0.0198) (0.0207) [0.706] [0.672] [0.776] [0.288] [0.223] [0.286] Covid19 epoch FE Yes Yes Yes Yes Yes Yes Month FE Yes Yes Yes Yes Yes Yes Court-in-epoch FE No Yes No Yes No Yes Staff per judge control No Yes No Yes No Yes Observations 552 552 552 552 552 552 R-squared 0.1660 0.5355 0.0008 0.9747 0.0269 0.5955 Notes: See notes under Table 2. Notably, we also find no evidence of an effect of the pandemic on the share of payments executed upon judicial enforcement. In particular, the pandemic and the accompanying economic downturn have, at least on average, evidently not resulted in an increased use of judicial enforcement as means to securing the execution of court-sanctioned payments stemming from the resolution of labor disputes. Finally, Table 5 shows the results for the effect of the pandemic on the composition of new filings with respect to enforcement and adjudication matters. The estimates reveal that, as a result of the pandemic, the share of new filings requiring judicial enforcement on average increased by 0.06, an effect equal to about 18 percent of the mean monthly share of newly filed enforcement cases between November to February in the Covid19 epoch or between March and October in the pre-Covid19 epoch. This is evidence that, even though the judicial enforcement outcomes do not seem to have been impacted, the pandemic has altered the composition of the demand for labor court services. As a consequence of the pandemic, there has been an increase in the demand for enforcement relative to adjudication. As parts (b) and (i) of Fig. 2 indicate, much of this composition effect can be attributed to a reduction in the overall demand for adjudication.Table 5 Effect on composition of new filings. Table 5 Outcome: Share enforcement new filings (1) (2) Post March × Covid19 epoch 0.0589*** 0.0581*** (0.0115) (0.0122) [<0.000] [<0.000] Covid19 epoch FE Yes Yes Month FE Yes Yes Court-in-epoch FE No Yes Staff per judge control No Yes Observations 552 552 R-squared 0.2178 0.6604 Notes: See notes under Table 2. 5.2 Month-by-month effect The results in Tables 2 through 5 show the average effect of the pandemic, but do not reveal whether, and if so how, the effect perhaps varied from month to month. Yet we know that the pace of the pandemic itself has varied over time and, moreover, that, after the pandemic's onset, the Brazilian labor courts adopted different operational responses at different points in time (see Fig. 1). To capture the resulting dynamics, we turn to investigating the results based on specification (2). We summarize our findings with the aid of Fig. 3 . Each part of Fig. 3 summarizes the results for a specific outcome variable, showing the month-by-month effects from November 2019 to October 2020. The omitted (comparison) effect is that for February, the month immediately preceding the onset of the pandemic. For each month, we display the point estimate and the corresponding 95-percent confidence interval computed using heteroscedasticity-robust standard errors clustered at the level of the regional labor court.Fig. 3 Summary of month-by-month effects. Notes: The figure shows the point estimates and the corresponding 95-percent confidence intervals based on OLS estimates of model (2) for different outcome variables. The omitted comparison month effect is that for February Fig. 3 Parts (a) and (b) of Fig. 3 show the month-by-month effect of the pandemic on court efficacy in adjudication. Part (a) illustrates that the adverse effect on the courts' capacity to meet the ongoing demand for adjudication was especially large, and in fact increased in terms of the absolute magnitude, over the initial months of the pandemic. The deleterious effect was strongest in the month of May, when the decrease in the clearance rate exceeded 0.4, an effect equal to about 40 percent of the average monthly clearance rate attained in the months from November to February in the Covid19 epoch or 35 percent of the average monthly clearance rate attained between March and October in the pre-Covid19 epoch. After May, as the courts gradually managed to regroup and introduced virtual hearings (see Section 2.2), the negative effect on the clearance rate weakened, even if it did not fully disappear. This is evidence that, after about half a year after the start of the pandemic, the Brazilian labor courts did eventually find a way to at least partly cope with the demand for adjudication. Part (b) of Fig. 3 traces out the month-by-month effect on the stock of pending adjudication cases per judge. Given the persistent and large negative effect on the clearance rate, which pushed the monthly clearance rate below the benchmark value of one, the pandemic resulted in persistent accumulation of unresolved cases. The effect on the increase in backlogs is largest for the final month covered by our data (October 2020), when the increase in the stock of pending adjudication cases per judge relative to February of the same year was equal to about 111 cases. This is an effect of the size of about 47 percent of the volume of pending adjudication cases per judge during the months from November to February in the Covid19 epoch or 43 percent of the volume of pending adjudication cases per judge during the months from March to October in the pre-Covid19 epoch. The documented effect on court backlogs is an indication that the pandemic will likely have a lasting detrimental impact on the ability of the Brazilian labor courts to administer justice in a timely manner. Parts (c) through (e) of Fig. 3 show the month-by-month effect of the pandemic on the modes of disposition of adjudication cases. Predictably, since its onset, the month-by-month effect of the pandemic has been a reduction in the share of cases resolved via trial hearings. The reduction was most pronounced in the months of April and May, when the share of cases resolved via hearings dropped by around 0.3, bringing the average share of cases resolved via trial hearings effectively to zero (the average monthly share of cases resolved via trial hearings was about 0.30 between November and February in the Covid19 epoch and about 0.33 between March and October in the pre-Covid19 epoch). Furthermore, the negative effect on the share of cases resolved via trial hearings has been a persistent one. Thus, the transition of courts to virtual hearings in the late spring alleviated, but did not eliminate, the deleterious effect of the pandemic on the ability of the courts to rely on hearings to resolve disputes. Based on the estimates in columns (4) through (6) of Table 3, the average effect of the pandemic on the share of cases disposed via settlement has been indistinguishable from zero. Part (d) of Fig. 3, however, shows evidence of intriguing dynamics with regard to month-by-month effect on the share of in-court settlements. Right after the onset of the pandemic, the effect was negative and very large: in April, for example, the pandemic resulted in a decrease in the share of adjudication cases settled in-court of about 0.17, an effect equal to approximately 48 percent of the mean monthly value of the share of cases settled in-court between November and February in the Covid19 epoch or 46 percent of the same value between March and October in the pre-Covid19 epoch. This pattern is consistent with the interpretation that the initial suspension of hearings and nation-wide restrictions on in-person interaction obstructed the exchange of information between the disputing parties, a process that normally facilitates settlement via the convergence of the disputing parties' views about the likely trial outcome (Boyd and Hoffman, 2013; Bielen et al., 2017, 2020). From June, however, the month-by-month effect on the share of cases settled in-court becomes positive, with the magnitude of the effect reaching about 18 percent of the average monthly share of cases settled in-court between November and February in the Covid19 epoch or between March and October in the pre-Covid19 epoch. That is, once the courts began to implement virtual hearing sessions, the pandemic's effect on the share of in-court settlements turned to positive. It appears, therefore, that the combination of the re-instituted possibility for court-facilitated exchange of information among the disputing parties and the judges, the awareness about the mounting court backlogs with corresponding increased prospects of court delays, and the unavoidable economic turmoil increased the relative attractiveness of settlement as means to resolution of labor disputes. On the other hand, congruent with the estimates of the average effect of the pandemic on the share of withdrawals, noted in the previous section, we find no evidence of an effect on this particular court outcome during any of the months. The pandemic has thus not noticeably altered the plaintiffs' incentives to altogether abandon their claims. The remaining parts of Fig. 3 summarize the results for the month-by-month effect of the pandemic on the enforcement outcomes (parts (f) through (h)) and the composition of incoming cases with respect to enforcement versus adjudication (part (i)). In line with the estimates of the average effect, reported in Table 4, we see little evidence of an effect of the pandemic on the efficacy of the enforcement aspects of labor court operations (parts (f) and (g)). In the month of April, there was a temporary positive effect on the clearance rate in enforcement. This rather peculiar effect, however, is driven primarily by a temporary drop in the number of newly filed enforcement cases in that month (see part (i) of Fig. 2), an artifact of the administrative chaos that accompanied the onset of the pandemic and impacted case registration processes across all labor courts. The month-by-month results also indicate that the pandemic temporarily and intermittently increased the reliance of litigants on judicial enforcement as means of executing court-ordered payments. The effect of the pandemic on the share of court-endorsed payments secured via judicial enforcement is positive and statistically significant (at five-percent level) in the months of June and August. The impact, however, dissipates by September (part (h)). In contrast, congruent with the results reported in Table 5, the impact of the pandemic on the composition of newly filed labor-court cases with respect to enforcement versus adjudication has been comparatively more long-lived (part (i)). Reflecting a drop in the overall demand for adjudication (see part (b) of Fig. 2), the pandemic resulted in a sizeable and enduring increase in the relative demand for enforcement versus adjudication. Last but not least, Fig. 3 illustrates that seven out of the nine court outcomes under consideration do not exhibit any unwanted pre-trends in the months prior to March. The share of cases resolved via hearings (part (c)) and the share of in-court settlements (part (d)) show some limited evidence of pre-trends. For the corresponding outcomes, our point estimates should thus be interpreted with some caution. All in all, however, the insights based on Fig. 3 complement the insights based on Fig. 2 in lending support for the use of the difference-in-difference approach in our setting. 5.3 Effect heterogeneity by pandemic intensity Tables 6 through 9 present the results on the heterogeneity of the effect of the pandemic by the intensity of the pandemic. In each of the tables, odd-numbered columns show the results when we measure the intensity of the pandemic using the rate of new Covid19 cases per 1000 people. Even-numbered columns display the results when we measure the intensity of the pandemic using the rate of Covid19-related deaths per 10,000 people.Table 6 Effect heterogeneity by pandemic intensity, efficacy in adjudication. Table 6 Outcome: Clearance rate (adj.) Outcome: Pending per judge (adj.) (1) (2) (3) (4) Post March × Covid19 epoch × New Covid19 cases (per 1 K) −0.0154 3.6297*** (0.0079) (0.8226) [0.045] [<0.000] Post March × Covid19 epoch × Covid19 deaths (per 10 K) −0.1053** 12.1539*** (0.0276) (2.5555) [0.001] [<0.000] Post March × Covid19 epoch −0.1670** −0.1290* 49.2408*** 51.4456*** (0.0506) (0.0454) (7.8229) (7.8418) [0.005] [0.009] [<0.000] [<0.000] Covid19 epoch FE Yes Yes Yes Yes Month FE Yes Yes Yes Yes Court-in-epoch FE Yes Yes Yes Yes Staff per judge control Yes Yes Yes Yes Observations 552 552 552 552 R-squared 0.4891 0.5159 0.9540 0.9528 Notes: The table presents OLS results based on the estimation of model (3). Post March is a dummy equal to one if observation is from the month of March or later in the applicable epoch. Covid19 epoch is a dummy equal to one if observation is from the Nov 2019-Oct 2020 period. New Covid19 cases and Covid19 deaths are measures of pandemic intensity (see Table A1 in the Appendix) that vary both across labor court regions and over time. Standard errors in parentheses are clustered at the level of regional labor court. In the brackets are p-values for the test of the null of no effect executed using wild bootstrap with 1000 replications. ***, **, and * denote statistical significance at the 0.1 %, 1%, and 5% level, respectively. We find, first and foremost, that the intensity of the pandemic has been an important moderating factor with regard to several court outcomes. Based on the results in Table 6, the detrimental effect of the pandemic on the ability of the courts to meet the ongoing demand for adjudication and on the stock of adjudicatory backlogs has been larger in the region-months characterized by greater pandemic intensity. For example, based on the estimates in columns (2) and (4) of Table 6, relative to the baseline effect of the pandemic, a one-standard-deviation increase in the number of Covid19-related deaths per 10,000 inhabitants is associated with an additional decrease in the clearance rate of 0.36 and an additional increase of 41 pending adjudication cases per judge. Thus, the severity of the health crisis has importantly shaped the effect of the pandemic on court efficacy in adjudication by exacerbating the pandemic's adverse impact. In addition, the intensity of the pandemic has played a role in shaping the effect of the pandemic on the court modes of case disposition in adjudication, and in particular on the use of in-court settlements (see Table 7 ). The effect of the pandemic on the share of in-court settlements is negative for the region-months that exhibit low pandemic intensity, but positive for the region-months that exhibit high pandemic intensity. Based on the estimates in columns (3) and (4) of Table 7, the effect of the pandemic on the share of in-court settlements turns from negative to positive as the number of new Covid19 cases per 1000 inhabitants reaches about 4.5 (about 117 percent of the mean monthly number of new Covid19 cases per 1000 inhabitants since March 2020) or, alternatively, the number of Covid19-related deaths per 10,000 inhabitants reaches about 1.2 (about 123 percent of the mean monthly number of Covid19-related deaths per 10,000 inhabitants since March 2020).Table 7 Effect heterogeneity by pandemic intensity, modes of disposition in adjudication. Table 7 Outcome: Share resolved in hearings Outcome: Share settled in-court Outcome: Share withdrawn (1) (2) (3) (4) (5) (6) Post March × Covid19 epoch × New Covid19 cases (per 1 K) −0.0002 0.0131*** −0.0002 (0.0025) (0.0023) (0.0003) [0.939] [<0.000] [0.380] Post March × Covid19 epoch × Covid19 deaths (per 10 K) −0.0184 0.0449*** −0.0008 (0.0128) (0.0097) (0.0016) [0.185] [0.001] [0.627] Post March × Covid19 epoch −0.1327*** −0.1172*** −0.0589*** −0.0519*** −0.0036 −0.0036 (0.0221) (0.0222) (0.0091) (0.0096) (0.0046) (0.0040) [<0.000] [<0.000] [<0.000] [<0.000] [0.431] [0.354] Covid19 epoch FE Yes Yes Yes Yes Yes Yes Month FE Yes Yes Yes Yes Yes Yes Court-in-epoch FE Yes Yes Yes Yes Yes Yes Staff per judge control Yes Yes Yes Yes Yes Yes Observations 552 552 552 552 552 552 R-squared 0.4458 0.4521 0.5987 0.5786 0.7641 0.7641 Notes: See notes under Table 6. Interestingly, the intensity of the pandemic appears to have played no role in moderating the effect of the pandemic on the share of cases resolved via trial hearings (columns (1) and (2) of Table 7). This finding is likely a reflection of the fact that, upon the onset of the pandemic, trial hearings came to a complete halt across all Brazilian labor courts at roughly the same time. Trial hearings eventually also resumed, albeit in a restricted (virtual) format, across all labor courts after the month of May (see Section 2.2). In this sense, the intensity of the pandemic per se has therefore not been a salient moderating factor. We also find no evidence of the importance of the intensity of the pandemic as a moderator of the effect of the pandemic on any of the judicial enforcement outcomes (see Table 8 ). More generally, congruent with the findings reported in Section 5.1, the effect of the pandemic on any of the judicial enforcement outcomes remains undetectable even upon allowing the effect of the pandemic to vary with pandemic intensity.Table 8 Effect heterogeneity by pandemic intensity, judicial enforcement. Table 8 Outcome: Clearance rate (enf.) Outcome: Pending per judge (enf.) Outcome: Share enforced payments (1) (2) (3) (4) (5) (6) Post March × Covid19 epoch × New Covid19 cases (per 1 K) −0.0133 −1.7954 0.0048 (0.0089) (2.9124) (0.0027) [0.114] [0.847] [0.060] Post March × Covid19 epoch × Covid19 deaths (per 10 K) −0.0435 −5.9170 0.0079 (0.0341) (9.7524) (0.0075) [0.193] [0.917] [0.274] Post March × Covid19 epoch 0.1139 0.1047 −5.0054 −6.1796 0.0079 0.0179 (0.1370) (0.1396) (8.5330) (8.0094) (0.0233) (0.0226) [0.407] [0.449] [0.550] [0.441] [0.731] [0.421] Covid19 epoch FE Yes Yes Yes Yes Yes Yes Month FE Yes Yes Yes Yes Yes Yes Court-in-epoch FE Yes Yes Yes Yes Yes Yes Staff per judge control Yes Yes Yes Yes Yes Yes Observations 552 552 552 552 552 552 R-squared 0.4909 0.4901 0.9748 0.9747 0.5742 0.5704 Notes: See notes under Table 6. The intensity of the pandemic, however, has shaped the effect of the pandemic on the composition of new filings. According to the estimates in Table 9 , a one-standard-deviation increase in the number of Covid19-related deaths per 10,000 inhabitants, or, alternatively, an equivalent increase in the number of new Covid19 infections per 1000 inhabitants, is associated with an additional increase in the share of new filings that necessitate judicial enforcement of about 0.06. Hence, all else equal, the relative demand for judicial enforcement has been greater in areas and time periods exhibiting greater pandemic intensity.Table 9 Effect heterogeneity by pandemic intensity, composition of new filings. Table 9 Outcome: Share enforcement new filings (1) (2) Post March × Covid19 epoch × New Covid19 cases (per 1 K) 0.0090*** (0.0018) [<0.000] Post March × Covid19 epoch × Covid19 deaths (per 10 K) 0.0346*** (0.0068) [<0.000] Post March × Covid19 epoch 0.0273* 0.0287* (0.0103) (0.0123) [0.011] [0.020] Covid19 epoch FE Yes Yes Month FE Yes Yes Court-in-epoch FE Yes Yes Staff per judge control Yes Yes Observations 552 552 R-squared 0.6259 0.6229 Notes: See notes under Table 6. 6 Summary and conclusion We have provided the first systematic empirical analysis of the consequences of the Covid19 pandemic for the performance of courts at accomplishing their primary function, the disposition of cases. Using a newly-assembled, monthly panel of Brazilian regional labor courts and employing a difference-in-difference approach, we have demonstrated, first, that the pandemic has had a very large and persistent deleterious impact on adjudicatory efficacy. Timing-wise, the adverse effects on adjudicatory efficacy were especially drastic in the first few months after the onset of the pandemic, when many court activities were altogether suspended. Upon the eventual introduction of virtual court hearings and other operational measures intended to facilitate adjudication, the drop in the clearance rate of adjudication cases decreased somewhat in absolute magnitude, but, at least by the end of our observation window, never vanished. More generally, the adverse effect on court efficacy has been largest in region-months where the rates of new Covid19 infections or Covid19-related deaths have been highest. Second, the pandemic has affected the modes through which the labor courts resolve disputes. As anticipated, following the suspension of in-person activities, including trial hearings, the share of cases resolved in hearings effectively dropped to zero. Concurrently, the share of cases resolved via in-court settlement fell as well. This is consistent with the interpretation that the restraints on in-person interaction and suspension of hearings limited the inter-party exchange of information about the cases, which in turn reduced the prospects of settlement. However, once the courts instituted virtual hearing sessions, as the prospects of court delays became apparent, and when a prolonged recession was clearly in sight, the effect of the pandemic on the share of in-court settlements turned to positive. Resonating with this explanation, we also find that the effect of the pandemic on the share of in-court settlements was positive in region-months characterized by the highest rate of new Covid19 infections or the largest Covid19-related death toll rate. The pandemic, however, did not affect the share of cases disposed as a result of withdrawals of already started lawsuit. Third, we find little evidence of an effect of the pandemic on judicial enforcement outcomes. In particular, our estimates reveal that the pandemic has not affected court efficacy in the context of judicial enforcement, even though the pandemic has increased the relative demand for enforcement versus adjudication. Because judicial enforcement of final decisions in Brazilian labor courts can be effort-intensive and, indeed, constitutes a sizeable portion of the courts' dockets, our findings indicate that the pandemic-induced transition of the Brazilian labor court judges toward a remote completion of the enforcement tasks was overall quite effective. We also find only limited evidence that the pandemic has increased the need for judicial enforcement as means of securing the execution of court-sanctioned payments. To the extent than such an effect is detectable, the effect has been intermittent and transitory. During the first eight months from the onset of the pandemic, we therefore do not see much indication that the losing parties (e.g. employers), who are mandated by the court to compensate the winning parties (e.g. employees), have, perhaps as a reaction to the economic recession, strategically chosen to avoid the execution of court-ordered payments. This particular result is perhaps the most uplifting of all findings that have emerged from our analysis. It suggests that, at least when it comes to the timing of received compensation, awarded on the basis of unlawful violation of labor contracts, the employees who are eligible for such compensation and who often stem from the most vulnerable segments of the Brazilian society have, on average, not really been adversely impacted by the pandemic. Overall, however, one may anticipate that the pandemic will continue to exert a lasting negative impact on the ability of the Brazilian labor courts to deliver justice, thereby exacerbating already considerable pre-existing socio-economic inequality. Our goal in this paper has been to offer an empirical analysis of the effect of the Covid19 pandemic on some of the most salient aspects of court operations. We have done so in the context of Brazilian labor justice, utilizing rich up-to-date court-level data. Future research will undoubtedly uncover many more aspects in which the pandemic has already impacted and will in the foreseeable future continue to impact the functioning of courts and the administration of justice both in Brazil and in other jurisdictions worldwide. Especially in the context of labor justice it would be pertinent to examine whether, and if so in what way, the pandemic has affected the composition of court dockets with respect to case complexity and the exact substance of labor-related claims. The resulting analysis would also help illuminate the extent to which the documented efficacy effects of the pandemic can be attributed to the operational responses of the courts versus any case composition effects. Investigation of such questions will require access to case-level data. At the same time, when even more recent court-level data become available, it will be important to provide updated estimates of the consequences of the pandemic for the outcomes explored in the present paper. Indeed, at the time of our writing, Brazil is already amidst a new devastating wave of the spread of the virus. In the current time, when the pandemic is nowhere close to under control, medical and natural-science research aimed at preventing, to the greatest extent possible, further loss of human life and securing adequate healthcare responses to the ravaging virus should remain the highest priority. For social scientists and other scholars, however, it will be important to enhance our understanding of the impact of the pandemic on key societal institutions, including the courts of law, and propose appropriate policy responses to help mitigate the associated rapidly rising social costs. Such research will necessarily require evidence-based insight of the type that we have striven to provide in the current paper. Declaration of Competing Interest None. Appendix Table A1 Variable definitions. Table A1Variable Definition Courts, adjudication  Clearance rate (adj.) The number of resolved adjudication cases during a month divided by the number of newly filed adjudication cases during the same month.  Pending per judge (adj.) The number of adjudication cases pending at the end of a month divided by the number of serving judges during the same month.  Share resolved in hearings The number of adjudication cases resolved in trial hearings during a month divided by the total number of resolved adjudication cases during the same month.  Share settled in-court The number of adjudication cases settled in-court during a month divided by the total number of resolved adjudication cases during the same month.  Share withdrawn The number of withdrawn adjudication cases during a month divided by the total number of resolved adjudication cases during the same month. Courts, enforcement  Clearance rate (enf.) The number of completed enforcement proceedings during a month divided by the number of newly initiated enforcement cases during the same month.  Pending per judge (enf.) The number of enforcement cases pending at the end of the month divided the number of serving judges during the same month.  Share enforced payments The value of payments executed upon completed judicial enforcement proceedings during a month divided by the total value of all executed payments during the same month. Courts, new filings composition  Share enforcement new filings The number of newly initiated enforcement cases during a month divided by the number of all new (adjudication and enforcement) case filing during the same month. Courts, staffing  Staff per judge The number of judicial support staff (judicial assistants and administrative staff) during a month divided by the number of serving judges during the same month. Pandemic intensity  New Covid19 cases (per 1,000 people) New Covid19 cases per capita in a given month, multiplied by 1,000.  Covid19 deaths (per 10,000 people) Covid19-related deaths per capita in a given month, multiplied by 10,000. Notes: The table provides the definitions of the outcome and select other variables used to generate the estimates shown in Table 1, Table 2, Table 3, Table 4, Table 5, Table 6, Table 7 and Fig. 3. Acknowledgement For helpful comments and suggestions, we thank an anonymous reviewer and Eric Helland, our editor. 1 See, for example, the RAND Corporation's "COVID19 and the Courts" virtual event (https://www.rand.org/events/2020/10/01.html). 2 See https://coronavirus.jhu.edu/data/mortality. 3 The information presented in this section draws heavily on parts of Section 2 in Castelliano et al. (2020b). 4 Because each labor court office consists of a titled and a substitute judgeship and since the allocation of cases between the two judgeship types within each office is random, the enforcement proceedings are not necessarily carried out by the same judge that adjudicated the original case. 5 Access to justice was further facilitated by the fact that many states permitted the re-opening of law offices soon after the initial March lockdown. In addition, many law offices swiftly transitioned to offering their services online. 6 The labor court regions normally coincide with Brazilian state borders. The exceptions are the 8th, 10th, 11th, and 14th labor court region, each of which extends over two states. In addition, the state of São Paulo comprises two labor court regions: the 2nd region (the city of São Paulo) and the 15th region (the remaining part of the state). 7 At the time of our conducting of this research and the writing up of the results, more recent monthly labor-court data are not (yet) available. 8 Because we view the state of São Paulo as a single labor court region and because the remaining labor court regions either coincide with state borders or precisely absorb two states (see note 6 above), the merger of the regional labor court data and Covid19 data is exact. 9 As a robustness check, we redefined the pre-Covid19 epoch as the November 2016-October 2017 period, the latest suitable alternative 12-month span between November and the following October for which monthly labor court data are available to us. (The November 2017-October 2018 period is not an appropriate choice for the purpose at hand because the end of year 2017 encompasses the start and implementation of a major reform.) The results obtained on the basis of this alternative definition of the pre-Covid19 epoch were both qualitatively and quantitatively very similar to the results reported in Section 5. ==== Refs References Baldwin Julie Marie Eassey John M. Brooke Erika J. Court operations during the COVID-19 pandemic Am. J. Crim. Justice 2020 10.1007/s12103-020-09553-1 forthcoming Beenstock Michael Haitovsky Yoel Does the appointment of judges increase the output of the judiciary? Int. Rev. Law Econ. 24 3 2004 351 369 Bełdowski Jarosław Dąbroś Łukasz Wojciechowski Wiktor Judges and court performance: a case study of district commercial courts in Poland Eur. J. Law Econ. 50 1 2020 171 201 Bielen Samantha Grajzl Peter Marneffe Wim Procedural events, judge characteristics, and the timing of settlement Int. Rev. Law Econ. 52 2017 97 110 Bielen Samantha Grajzl Peter Marneffe Wim The resolution process and the timing of settlement of medical malpractice claims Health Econ. Policy Law 15 4 2020 509 529 30994084 Boyd Christina L. Hoffman David A. Litigating toward settlement J. Law Econ. Organ. 29 4 2013 898 929 Buscaglia Edgardo Ulen Thomas S. A quantitative assessment of the efficiency of the judicial sector in Latin America Int. Rev. Law Econ. 17 2 1997 275 291 Castelliano Caio Grajzl Peter Alves Andre Watanabe Eduardo Adjudication forums, specialization, and case duration: evidence from brazilian federal courts Justice Syst. J. 2020 10.1080/0098261X.2020.1854905 forthcoming Castelliano Caio Grajzl Peter Guimaraes Tomas A. Alves Andre "Judicial enforcement and caseload heterogeneity: theory and evidence from Brazil Working Paper, Presented at the EMLE Midterm Meeting University of Hamburg, Germany 2020 CEPEJ - The European Commission for the Efficiency of Justice National Judiciaries’ COVID-19 Emergency Measures of COE Member States 2020 https://www.coe.int/en/web/cepej/national-judiciaries-covid-19-emergency-measures-of-coe-member-states CEPEJ - The European Commission for the Efficiency of Justice European Judicial Systems–Edition 2020 (2018 data): Efficiency and Quality of Justice 2020 Council of Europe Strasbourg, France Chemin Matthieu. The impact of the judiciary on entrepreneurship: evaluation of Pakistan’s’ Access to justice programme’ J. Public Econ. 93 1–2 2009 114 125 Christensen Robert K. Szmer John Examining the efficiency of the U.S. courts of appeals: pathologies and prescriptions Int. Rev. Law Econ. 32 1 2012 30 37 CNJ - Conselho Nacional de Justiça Relatório Justiça Em Números 2020: Ano-base 2019 2020 Brasil: Conselho Nacional de Justiça Brasilia Di Vita Giuseppe. Factors determining the duration of legal disputes: an empirical analysis with Micro data J. Inst. Theor. Econ. 168 4 2012 563 587 Dimitrova-Grajzl Valentina Grajzl Peter Sustersic Janez Zajc Katarina Court output, judicial staffing, and the demand for court services: evidence from slovenian courts of first instance Int. Rev. Law Econ. 32 1 2012 19 29 Dimitrova-Grajzl Valentina Grajzl Peter Zajc Katarina Understanding modes of civil case disposition: evidence from slovenian courts J. Comp. Econ. 42 4 2014 924 939 EBRD - European Bank for Reconstruction and Development Law in Transition 2014 2014 European Bank for Reconstruction and Development. London, UK Engstrom David Freeman Post-COVID courts UCLA Law Review Discourse 68 2020 246 267 Galanter Marc. The vanishing trial: an examination of trials and related matters in federal and state courts J. Empir. Leg. Stud. 1 3 2004 459 570 Grajzl Peter Silwal Shikha Multi-court judging and judicial productivity in a career judiciary: evidence from Nepal Int. Rev. Law Econ. 61 2020 105888 Marciano Alain Melcarne Alessandro Ramello Giovanni B. The economic importance of judicial institutions, their performance and the proper way to measure them J. Inst. Econ. 15 1 2019 81 98 Matyas David Wills Peter Dewitt Barry Imagining Resilient Courts: From COVID to The Future of Canada’s Judicial System SSRN Working Paper No.3778869 2021 McIntyre Joe Olijnyk Anna Pender Kieran Civil courts and COVID-19: challenges and opportunities in Australia Altern. Law J. 45 3 2020 195 201 Pistor Katharina Law in the Time of COVID-19 2020 Columbia Law School. New York, NY Puddister Kate Small Tamara A. Trial by zoom? The response to COVID-19 by Canada’s courts Can. J. Political Sci. 53 2020 373 377 Roodman David Nielsen MortenØrregaard MacKinnon James G. Webb Matthew D. Fast and wild: bootstrap inference in Stata Using Boottest Stata J. 19 2019 4 60 Rosales-López Virginia Economics of court performance: an empirical analysis Eur. J. Law Econ. 25 3 2008 231 251 Sourdin Tania Zeleznikow John Courts, mediation and COVID-19 Australian Bus. Law Review 2020 forthcoming TST - Tribunal Superior do Trabalho Relatório Geral Da Justiça Do Trabalho 2020 Brasil: Secretaria-Geral da Presidência do TST Brasilia Voigt Stefan. Determinants of judicial efficiency: a survey Eur. J. Law Econ. 42 2 2016 183 208 Warner Randall H. Judging in a time of COVID Fam. Court Rev. 58 4 2020 965 967

0

PMC9754149

NO-CC CODE

2022-12-16 23:26:27

no

Int Rev Law Econ. 2021 Jun 31; 66:105989

utf-8

Int Rev Law Econ

10.1016/j.irle.2021.105989

oa_other

==== Front Int Rev Law Econ Int Rev Law Econ International Review of Law and Economics 0144-8188 1873-6394 Elsevier Inc. S0144-8188(21)00013-2 10.1016/j.irle.2021.105989 105989 Article How has the Covid19 pandemic impacted the courts of law? Evidence from Brazil Castelliano Caio a Grajzl Peter bc⁎ Watanabe Eduardo d a University of Brasilia, Brazil b Washington and Lee University, USA c CESifo, Germany d University of Brasilia, Brazil ⁎ Corresponding author at: Department of Economics, The Williams School of Commerce, Economics and Politics, Washington and Lee University, 204 West Washington St., Lexington, VA, 24450, USA. 31 3 2021 6 2021 31 3 2021 66 105989105989 1 3 2021 28 3 2021 29 3 2021 © 2021 Elsevier Inc. All rights reserved. 2021 Elsevier Inc. Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. We provide empirical insight into the consequences of the Covid19 pandemic for the administration of justice. Drawing on a comprehensive monthly panel of Brazilian labor courts and using a difference-in-difference approach, we show that the pandemic has had a large and persistent deleterious effect on adjudicatory efficacy, leading to a massive decrease in the clearance rate and an increase in court backlogs. The pandemic has affected how courts dispose adjudication cases, expectedly causing a plummeting in the share of disputes resolved via trial hearings and, less predictably, exerting a temporally non-linear effect on the share of in-court settlements. Notably, we find no evidence of an effect of the pandemic on efficacy in enforcement. Although the pandemic led to an increase in the share of new filings requiring enforcement, any effect on the relative use of enforcement to execute court-ordered payments has been intermittent and temporary. The intensity of the pandemic has been an important moderating factor. Keywords Covid19 Courts Brazil Labor justice Adjudication Enforcement ==== Body pmc1 Introduction The Covid19 pandemic has fundamentally impacted virtually every facet of our existence, wreaking havoc in healthcare systems, leading to a massive death toll, and causing a profound economic and social disruption. Unsurprisingly, the functioning of judicial systems has been impacted as well. Across jurisdictions worldwide, lockdowns, self-isolation, and restrictions on the population's movement and assembly have affected the ability of the courts to perform their function. Summarizing the state of affairs at the onset of the pandemic, the European Commission for the Efficiency of Justice, for example, noted that “the courts are facing numerous challenges to remain operational due to lack of personnel, hearings are being cancelled, and access to justice is temporarily limited” (CEPEJ, 2020a). Yet despite the agreement among policymakers, court administrators and scholars about the pressing nature of the problem1 , we are aware of no systematic quantitative study characterizing the impact of the pandemic on the output and performance of the courts of law. In this paper, we provide the first full-fledged empirical inquiry into the effect of the Covid19 pandemic on the success of courts at performing their primary function, the resolution of disputes and disposition of cases. To this end, we draw on comprehensive data on the activity of labor courts in Brazil. In Brazil, the impact of the pandemic has been especially severe.2 At the same time, the court system as a whole, and labor courts in particular, have been subject to a series of pandemic-instigated policy responses that have fundamentally affected their operations. Brazilian labor courts possess jurisdiction over all labor law-related disputes, with the overarching majority of filed claims pertaining to infringements of employee-employer agreements, especially with regard to unlawful termination of employment and unpaid compensation. As such, labor courts constitute a crucial component of the Brazilian judicial system. Moreover, empirical insight into the effect of the pandemic on the administration of Brazilian labor justice provides valuable clues about the effects of the pandemic on the courts of law in other jurisdictions worldwide. Our data are unique in several respects. First, the data cover all Brazilian regional labor courts observed at a monthly level, a time frequency that is higher than that normally encountered in the empirical literature on court administration. Second, we observe the regional labor courts both in the era prior to and after the onset of the pandemic, a feature of the data that aids the assessment of the impact of the pandemic. Third, our dataset encompasses the indicators of court activity with regard to both adjudication and enforcement. Unlike the resolution of adjudication cases, the disposition of enforcement cases does not entail the conduct of hearings. Nevertheless, the enforcement of final decisions (res judicatae) requires judicial effort, represents a very significant share of dockets of the Brazilian labor courts, and, when required, constitutes a critical step in the process of restoration of justice for the prevailing party. Utilizing the resulting panel dataset, in which a unit of observation is a regional labor court in a given month, we are therefore able to offer an in-depth empirical glimpse into how the pandemic has affected the administration of justice in one prominent legal system. The assessment of the consequences of the pandemic, however, presents an empirical challenge. A naïve pre- versus post-pandemic comparison of court outcomes, for example, will not adequately capture the effect of the pandemic when court activity is subject to inherent temporal trends. We therefore adopt a difference-in-difference (DD) approach. To construct the missing counterfactuals depicting the changes in the labor court outcomes in the absence of the pandemic, we rely on the changes in the outcomes of the same set of courts observed during the pre-Covid19 epoch. That is, prior to March―the month of the onset of the pandemic in the year 2020―the trends in the outcomes in the Brazilian regional labor courts during the Covid19 epoch, defined as the 12-month span from November 2019 to October 2020 (the latest month of data availability), closely resemble those registered one year earlier. This feature renders the same set of regional labor courts that we observe also during the pre-Covid19 epoch, defined as the 12-month span from November 2018 to October 2019, a suitable control group for purposes of our analysis. Our DD estimates show that the Covid19 pandemic, above all, drastically adversely affected the adjudicative efficacy of Brazilian labor courts. We find that the pandemic-caused average drop in the monthly clearance rate in adjudication is of the size of about one-fifth of the mean monthly clearance rate during the pre-pandemic era. Consequently, the pandemic led to a surge in court backlogs in adjudication, with the estimated increase in the number of pending cases per judge equal to about a quarter of the mean monthly pending cases per judge recorded during the pre-pandemic era. Thus, the pandemic is expected to exert a lasting deleterious effect on the ability of the courts to deliver justice in due time. At the same time, the pandemic has impacted the modes with which the courts dispose adjudication cases. As one would have expected, the share of adjudication cases disposed via trial hearings decreased sharply as the courts implemented an initial ban on all in-person court activities. The pandemic, however, has at least on average not affected the share of adjudication cases disposed via in-court settlement or lawsuit withdrawal. Interestingly, we also do not find evidence of an impact of the pandemic on court outcomes in enforcement. Neither the clearance rate in enforcement nor the volume of pending enforcement cases per judge exhibit a detectable change as a result of the pandemic. Similarly, even though the pandemic resulted in an increase in the relative demand for judicial enforcement versus adjudication, we do not find evidence that the pandemic altered the relative reliance on judicial enforcement as means of securing the execution of court-sanctioned payments among the disputing parties. As we emphasize in interpreting our results, this last finding is an especially important and, in many respects, a reassuring one. At least during our observation window, the pandemic has on average not increased the need for judicial enforcement that could have conceivably arisen amidst the pandemic-instigated economic downturn, a period when the losing parties pressed for cash may be tempted to purposely avoid the execution of court-ordered damage and compensation payments. In addition to offering the baseline DD estimates of the average effect of the pandemic, we provide two further sets of estimates. We first explore an event-study approach to provide insight into the month-by-month effect of the pandemic. The response of the Brazilian labor courts to the pandemic has evolved with the evolution of the pandemic itself. Intuitively, one would therefore expect that the effects of the pandemic on court outcomes varied over time. As we demonstrate, the estimates based on a dynamic specification indeed reveal important insights into the temporal effects of the pandemic that would remain hidden if one focused solely on the static DD estimates. Our analysis shows, for example, that the negative effects on the clearance rate in adjudication were largest in the first few months after the start of the pandemic, but gradually diminished in magnitude as time unfolded and the courts implemented various response measures. Similarly, while the static DD estimates do not show an effect of the pandemic on the share of in-court settlements, the estimates based on the dynamic specification uncover that the initial effect on the share of in-court settlements was, in fact, negative, while the effect several months into the pandemic turned to positive. We interpret these findings in light of the existing theories of litigation. Finally, we investigate the heterogeneity of the effect of the pandemic with respect to the intensity of the pandemic. Since its onset, the Covid19 pandemic has in Brazil exhibited considerable geographic and temporal variation with regard to its severity, with different regions adopting different response measures at different points in time. The specificities of local conditions at a given point in time plausibly influenced the functioning of the courts in that locality at that point in time. To measure the intensity of the pandemic, we utilize official Brazilian monthly regional data on new Covid19 infections and Covid19-related deaths. We then estimate a specification where the DD effect of the pandemic is allowed to vary with pandemic intensity. Our findings suggest that the effect of the pandemic on court outcomes exhibits important heterogeneity with respect to pandemic intensity. We find, for instance, that the adverse effect of the pandemic on adjudicatory efficacy has been larger for those labor court region-months that have exhibited stronger pandemic intensity. Similarly, the average effect of the pandemic on the share of in-court settlements is positive in the labor court region-months characterized by a higher incidence of new infections or a greater virus death toll. But we once more do not find evidence of an effect of the pandemic on the core judicial enforcement outcomes. Our paper adds to and links two primary strands of literature. On the one hand, we contribute to the emerging literature on the consequences of the Covid19 pandemic for the courts of law. The Covid19 crisis has stimulated a global policy and academic debate about the numerous ways in which the pandemic has already affected, and will likely still affect, the operations of courts and the delivery of justice (e.g., McIntyre et al., 2020; Baldwin et al., 2020; Puddister and Small, 2020; Warner, 2020; Sourdin and Zeleznikow, 2020; Engstrom, 2020; Pistor, 2020; Matyas et al., 2021). The existing contributions on the topic, however, have been primarily descriptive in character. In particular, the research has not illuminated the impact of the pandemic on the administration of justice using comprehensive court data and rigorous quantitative analysis. To the best of our knowledge, our paper is the first to accomplish this task. At the same time, we contribute to the growing empirical literature on the administration of justice and the functioning of courts. Motivated by the ubiquity of court delays and the corresponding social costs incurred by numerous jurisdictions worldwide, an important subset of this literature has focused on the determinants of the efficacy of courts at performing their core role, the disposition of cases (e.g., Buscaglia and Ulen, 1997; Beenstock and Haitovsky, 2004; Rosales-López, 2008; Dimitrova-Grajzl et al., 2012; Di Vita, 2012; Chemin, 2009; Christensen and Szmer, 2012; Voigt, 2016; Marciano et al., 2019; Bełdowski et al., 2020; Grajzl and Silwal, 2020; Castelliano et al., 2020a). A related subset of research has investigated court-level factors affecting the use of different modes of court case disposition (e.g., Galanter, 2004; Dimitrova-Grajzl et al., 2014). In contrast, judicial enforcement of final verdicts, a distinct and often especially vital facet of court activity, has received much less attention (EBRD, 2014; Castelliano et al., 2020b). Our paper advances the corresponding literature by empirically examining the impact of the Covid19 pandemic on all three of the above-noted aspects of court output and operations. The rest of the paper is organized as follows. Section 2 provides an overview of the Brazilian labor courts and their response to the pandemic. Section 3 introduces our data. Section 4 develops our empirical approach. Sections 5 presents and discusses the results. The final section concludes. 2 Background 2.1 Brazilian labor justice and proceedings Labor justice, administered in the labor courts, is a key pillar of the Brazilian judicial system.3 In 2019, for example, new labor-court case filings represented about eleven percent of all new case filings in Brazilian (state and federal) courts (CNJ, 2020). Labor courts are specialized courts, with labor court judges following a career track that is separate from that followed by other federal and state judges. Brazil is divided into 24 labor court regions. Each labor court region is in turn divided into districts, with each district featuring one or more labor court offices. Each labor court region, however, has a single second-instance labor court, referred to as the regional labor court (tribunal regional do trabalho). In addition to adjudicating appeals to first-instance decisions, the regional labor court exercises administrative authority over the first-instance tribunals in its region. Jurisdictional rules require that labor law-related disputes be adjudicated at a first-instance court office with jurisdiction over the geographic area of the dispute's origination. In districts with more than one labor court office, a newly-filed case is allocated to a particular office using a system of computerized random assignment. Within any labor court office, cases are then allocated between a titled judgeship and a substitute judgeship using an analogous procedure (see Castelliano et al., 2020a). The institutional division of court offices into titled and substitute judgeships is intended to limit the possibilities of the litigants to engage in judicial forum shopping and, at the same time, operationalizes the Brazilian system of judicial career advancement where all new judges are first employed as substitute judges. The overwhelming majority (more than 99 percent) of cases adjudicated in Brazilian labor courts are employee claims stemming from alleged violation of employment contracts in the private sector. Among those, the most commonly brought-up subject issues are the termination of employment and overdue wages. Other frequent subject issues pertain to employer contributions into the public severance indemnity fund, severance payment, overtime wages, premiums for high-risk work, as well as compensation for pain and suffering (TST, 2020). Brazilian labor courts dispose cases in both the adjudication (conhecimento) and the enforcement (execução) stage of court proceedings, an important distinction emphasized by the Brazilian civil procedure and meticulously tracked by official court statistics. Specifically, a case filed at a first-instance labor court that is not settled, withdrawn, or dismissed on procedural grounds is eventually resolved via a court decision. To reach a decision, the adjudicating judge conducts hearings, interviews witnesses, and examines facts. The disputing parties may appeal at the second and higher instance, until the decision attains the status of a final decision. Often, the final decision specifies a monetary transfer from the losing to the winning party. For example, an employer who is found to have unlawfully terminated an employee's employment contract is expected to compensate the employee for the lost wages and any other damages. In those instances, the court orders the losing party to execute the payment to the benefit of the prevailing party. But the losing party sometimes does not comply. In the event of such non-compliance, the winning party must initiate separate judicial enforcement proceedings at the first-instance labor court office that adjudicated the original dispute.4 To enforce court-mandated payments, labor-court judges resort to a variety of patrimonial constrictions, the implementation of all of which requires judicial attention and time (see Castelliano et al., 2020b). Enforcement cases hence constitute a substantial portion of the labor courts' workload and facilitate the transfer of a very significant sum of awarded compensation payments. For example, in year 2019, the aggregate value of payments secured upon the completion of enforcement proceedings was more than three times as large the value of voluntary payments following the final court decisions (TST, 2020). 2.2 Operational response of labor courts to the pandemic In Brazil, as in many other countries, the first instances of Covid19 infections were officially registered in March 2020. Upon the call of the President of the Republic, the Brazilian National Congress formally recognized the state of public calamity on March 20. Even prior to that, on March 18, the National Council of Labor Justice decreed the suspension of all in-person services, and in particular the conduct of trial hearings, effective from March 19. The courts, including the labor courts, were to continue in an uninterrupted fashion only with the provision of essential services (such as information-technology support, facility security, and payroll). Importantly, all procedural deadlines related to case processing (for example, to file an appeal or include new evidence) were suspended until June 14, 2020. From March 19, the judges and their support staff were authorized to carry out remotely all substantive tasks, including the preparation of judgments and the execution of administrative duties. The resulting provision enabled the judges to implement remotely and via virtual sessions all key decision-making activities, with the exception of those that inherently rely on the conduct of trial hearings. The abrupt switch to remote work undoubtedly led to initial disorganization and difficulties. There was, however, also an anticipation that the judges and their staff would sooner or later adapt to the new work format. At the same time the courts continued to accept new filings. In Brazilian labor courts the overwhelming majority of cases had already been filed electronically even prior to the start of the pandemic. For example, in the year 2019, as many as 99 percent of all new labor cases were filed electronically. Therefore, at least when it comes to the logistics of the initiation of new filings, the onset of the pandemic was not expected to radically curtail access to labor justice.5 In addition, an executive order of the President of the Republic, valid from March 22 until July 19, enabled the employers to implement remote work, use individual and collective vacation days, as well defer payments into the public severance indemnity fund with the express aim of avoiding instances of breach of labor contracts. These measures naturally mitigated the demand for labor justice in anticipation of the pandemic-caused economic disruption. Once the inevitability of the prolonged nature of the pandemic became apparent and upon the passage of supporting regulations, on May 5, the National Council of Labor Justice authorized the courts to conduct virtual trial hearings. To take into consideration the local differences in both the intensity of the pandemic and the state and municipal government responses to it, the administrative and technological implementation of the conduct of virtual hearings was delegated to the regional labor courts themselves. The pace of the implementation of virtual hearings was therefore likely not uniform across labor courts and in particular across individual judges. Many judges, but also clients and their attorneys, certainly grappled with the usage of new technology. In addition to enabling the conduct of virtual hearings, the May 5 resolution by the National Council of Labor Justice re-instated the validity of the procedural deadlines that had been suspended since March 19. The judges, however, were granted discretion with regard to the application of the procedural deadlines pertinent to specific cases based on the epidemiological conditions relevant to each case. Finally, the resolution established criminal liability for the execution of any court operations that could potentially contribute to the spread of the virus. That is, the courts themselves were not to, and did not, become a contributor to the spread of the virus. On June 16, the National Council of Labor Justice formally allowed for a gradual reestablishment of in-person court activities, including the conduct of hearings, subject to the adoption of appropriate safety protocols by the courts. Much like in the case of the prior introduction of the possibility to carry out virtual hearings, the implementation of the eventual return to in-person hearings was entrusted to the regional labor courts themselves. Given the raging pandemic, the courts and especially the judges adopted a very cautious approach. Because all Brazilian federal judges, including labor court judges, are civil servants with lifetime tenure, the judges have not felt the pressure to physically appear in their formal office spaces to do their work. At the same time, the vast majority of the judges are proficient in the use of required information technology and managed to adapt to working from home. Thus, in practice, the physical facilities of nearly all courts, including the labor courts, remained closed to the public and the courts have continued to conduct their operations remotely, in a virtual format, even after June. Fig. 1 summarizes the key events pertinent to the functioning of the labor courts from the onset of the pandemic until October 2020, the last month of coverage in our data.Fig. 1 The timeline of the key operational responses of the labor courts to the pandemic. Fig. 1 3 Data The source of our data on regional labor courts is a database compiled by the Brazilian Superior Labor Court. There are 24 labor court regions in Brazil. For purposes of the analysis, we combine two labor court regions, which together encompass the state of São Paulo, into a single labor court region. The combining of these two labor court regions is necessary because consistent monthly Covid19-related data that we employ in a subset of our analysis are available only at the level of the state of São Paulo as a whole.6 For each of the correspondingly-defined 23 regional labor courts, we observe monthly data on court staffing and court activity over a 24-month span between November 2018 and October 2020.7 We split the resulting time span into two contiguous, non-overlapping 12-month subperiods. The first, from November 2019 to October 2020, includes the onset of the Covid19 pandemic in March 2020. We refer to this period as the Covid19 epoch. The second subperiod, from November 2018 and October 2019, covers the exact same months as the Covid19 epoch, but occurring one year earlier when the labor courts were not subject to any noteworthy shocks or legislative changes. We refer to this second subperiod as the pre-Covid19 epoch. As we clarify in Section 4 below, the resulting two-epoch structure of our data facilitates the estimation of the effect of the Covid19 pandemic on labor court outcomes. During each of the two epochs we for each labor court in every month observe outcomes indicative of the extent of court efficacy in the context of both adjudication and enforcement. Specifically, for each of the two types of proceedings, we observe the monthly number of resolved cases and the number of newly filed cases. For purposes of empirical analysis, we divide the number of resolved cases of a given type (adjudication or enforcement) by the number of newly filed cases of the same type. The resulting case type-specific clearance rate is an indicator of the ability of a court to meet the demand for its services in the pertinent domain. As such, clearance rate is a core and commonly utilized measure of court efficacy (see, e.g., CEPEJ, 2020b; Voigt, 2016). A value of the clearance rate greater than one indicates that the court is able to both meet the ongoing demand and reduce existing backlogs. In contrast, a value of the clearance rate smaller than one implies that the court is contributing to the accumulation of case backlogs. For both adjudication and enforcement cases, we also observe the total number of cases still pending at the end of every month at every court. For each court, we divide the total number of pending adjudication and enforcement cases, respectively, with the number of judges serving at the court during the applicable month. The resulting end-of-month number of pending adjudication and enforcement cases per judge, respectively, are direct measures of court backlogs. The combination of the clearance rate and the number of pending cases per judge in adjudication and enforcement, respectively, thus allows us to investigate the impact of the pandemic on court efficacy in adjudication and enforcement. At the adjudication stage, we further observe the monthly number of cases resolved via trial hearings, the number of in-court settlements, and the number of withdrawals. For every court in every month, we divide each of these three variables with the total volume of disposed adjudication cases. The share of cases resolved via trial hearings, the share of case settled in court, and the share of withdrawals are then measures of court output with regard to modes of case disposition. We use thus-defined outcomes to explore the consequences of the pandemic for the courts' modes of disposition of adjudication cases. For every court in each month, we observe the total value of all payments executed as a consequence of court proceedings and the value of payments executed only upon completion of the enforcement proceedings. In general, labor court-sanctioned payments are executed for one of three distinct reasons: a spontaneous transfer that is executed voluntarily by the losing party after a court verdict; an in-court settlement-based or conciliation-induced payment that occurs at any stage of the adjudication or enforcement proceedings; or a payment secured upon the completion of judicial enforcement proceedings if the losing party fails to comply with the final decision. We calculate the monthly share of the value of all executed payments that occur as a consequence of enforcement. The resulting measure is indicative of the relative importance of judicial enforcement for the execution of court-sanctioned payments. Given the decline in the economic activity as a consequence of the pandemic and the corresponding liquidity pressures faced by many employers, we want to examine whether the pandemic has resulted in greater reliance on judicial enforcement. Such a scenario could arise if cash-stripped employers, found to had violated the Brazilian labor law, perhaps purposefully chose to avoid voluntarily executing court-ordered damage and compensation payments. Finally, using the information on the volume of new filings of adjudication and enforcement cases at each court during every month, we compute the share of newly filed enforcement cases in all (adjudication and enforcement combined) newly filed cases. The resultant variable, a measure of the composition of new filings, allows us to gauge whether, and if so to what extent, the pandemic and the ensuing economic crisis have altered the balance between the demand for enforcement versus demand for adjudication. The data, however, do not allow us to observe the temporal evolution of the exact composition of newly-filed and processed adjudication and enforcement cases, respectively. Consequently, we are unable to ascertain, for example, to what extent any pandemic-induced changes in labor-court efficacy in adjudication and enforcement, respectively, are purely due to the response of the courts per se versus any changes in the complexity of the underlying cases or, when it comes to the clearance rate, perhaps even the strategic behavior of the litigants. Thus, while operational responses of the courts alone have undoubtedly been of central importance, our results should be interpreted as amalgamating multiple mechanisms. We merge the data for the 23 above-defined labor courts with the official monthly data on the incidence of new Covid19 cases and Covid19-related deaths, recorded at the geographic level of labor court regions and normalized using Brazilian census population data.8 This gives us monthly per-capita measures of the intensity of the Covid19 pandemic at the level of each labor court region. From March 2020 onwards, the resulting variables vary both from month to month and across the labor court regions. Table 1 shows the basic descriptive statistics for the outcome and selected additional variables that we use in the analysis for both the Covid19 epoch (part A) and the pre-Covid19 epoch (part B). Table A1 in the Appendix provides the corresponding variable definitions.Table 1 Descriptive statistics. Table 1 Part 1: Covid19 epoch (Nov 2019-Oct 2020) Part A1: Nov 2019-Feb 2020 Part A2: Mar 2020-Oct 2020 Obs. Mean S.D. Obs. Mean S.D. Courts, adjudication  Clearance rate (adj.) 92 1.0871 0.2145 184 0.8662 0.2553  Pending per judge (adj.) 92 235.3 94.3 184 260.7 102.3  Share resolved in hearings 92 0.2952 0.0877 184 0.1808 0.1003  Share settled 92 0.3511 0.0754 184 0.3659 0.1240  Share withdrawn 92 0.0428 0.0305 184 0.0418 0.0201 Courts, enforcement  Clearance rate (enf.) 92 1.2727 0.6163 184 1.3766 0.5462  Pending per judge (enf.) 92 611.2 214.0 184 599.3 239.8  Share enforced payments 92 0.4377 0.1338 184 0.4689 0.1551 Courts, staffing  Staff per judge 92 5.12 1.05 184 5.06 1.03 Courts, new filings composition  Share enforcement new filings 92 0.3147 0.0740 184 0.3831 0.0921 Pandemic intensity  New Covid19 cases per 1000 people 92 0 0 184 3.84 3.41  Covid19 deaths per 10,000 people 92 0 0 184 0.97 0.84 Part B: Pre-Covid19 epoch (Nov 2018-Oct 2019) Part B1: Nov 2018-Feb 2019 Part B2: Mar 2019-Oct 2019 Obs. Mean S.D. Obs. Mean S.D. Courts, adjudication  Clearance rate (adj.) 92 1.1829 0.2142 184 1.2120 0.1683  Pending per judge (adj.) 92 306.6 139.8 184 260.3 118.1  Share resolved in hearings 92 0.2815 0.1251 184 0.3305 0.0831  Share settled 92 0.3422 0.0756 184 0.3687 0.0703  Share withdrawn 92 0.0385 0.0216 184 0.0410 0.0256 Courts, enforcement  Clearance rate (enf.) 92 0.9500 0.3241 184 0.9846 0.4896  Pending per judge (enf.) 92 608.4 217.1 184 602.7 221.6  Share enforced payments 92 0.4613 0.1387 184 0.4607 0.1295 Courts, new filings compositon  Share enforcement new filings 92 0.3216 0.0560 184 0.3291 0.0532  Courts, staffing  Staff per judge 92 5.26 1.12 184 5.11 1.04 Notes: The table presents descriptive statistics for the eight outcome variables used in the empirical analysis, a control (staff per judge), and moderating variables (on pandemic intensity). Observation is a labor court (or, equivalently, labor court region) in a given month. 4 Empirical approach To obtain an initial glimpse into the consequences of the pandemic for court outcomes, one could imagine pursuing two simple approaches. Under one approach, one might contrast the outcomes post March with the outcomes prior to March during the Covid19 epoch, that is, compare the mean for the outcomes of interest in part A2 of Table 1 with the mean for the same outcomes in part A1 of Table 1. The resulting approach, however, does not take into account the trends in the data. For instance, the pre-March period subsumes the holiday-season months of December and January, when court activity naturally slows down every year. The post-March versus pre-March comparison alone would therefore unlikely yield a compelling estimate of the effect of the pandemic. Alternatively, one might contrast the post-March outcomes from the Covid19 epoch with the post-March outcomes from the pre-Covid19 epoch, that is, compare the mean for the outcome in part A2 of Table 1 with the mean for the same outcomes in part B2 of Table 1. Yet the comparison of the post-March outcomes from the Covid19 epoch with the post-March outcomes from the pre-Covid19 epoch does not address the concern that the Covid19 and pre-Covid19 epochs plausibly differ in unobserved ways, which confounds the estimate of the effect of the pandemic. To address the deficiencies inherent in the two naïve approaches described above while at the same time combining their intuitively-appealing features, we use a difference-in-difference approach and exploit the exogenous nature of the pandemic. We first posit the following general model:(1) ycet=α+βPostt×Covide+γCovide+λt+μce+δSPJcet+εcet, where c denotes labor court, e epoch (pre-Covid19 or Covid19), and t month (from November to October the following year). ycet is one of the eight outcome variables, listed in the first eight entries in Table 1 or Table A1 in the Appendix. Postt is a dummy equal to 1 if the observation is from the month of March or later. Covide is a dummy equal to 1 if the observation is from the Covid19 epoch and 0 if it is from the pre-Covid19 epoch. The month fixed effect λt fully absorbs the timing of the observation relative to the start of the Covid19 pandemic, rendering a separate inclusion of Postt on the right-hand side of (1) redundant. μce is the labor court-in-epoch fixed effect, which absorbs the time-invariant average impact on the outcome under consideration of each labor court during each epoch. SPJcet is the staff per judge control that varies across labor courts, epoch, and months. εcet is the error term. The coefficient of interest in expression (1) is β, the difference-in-difference (DD) estimate of the impact of the Covid19 pandemic―our treatment of interest―on the court outcome under consideration. More precisely, β captures the difference between the post-March versus pre-March change in the court outcome under consideration during the Covid19 epoch and the analogous change during the pre-Covid19 epoch. That is, the first difference contrasts the post-March court outcome with the pre-March outcome in the Covid19 epoch, our treated group. However, we do not observe how the court outcome under consideration would have looked like after March 2020 had the pandemic never occurred. To construct the pertinent counterfactual, we use the change in the post-March versus pre-March outcome during the pre-Covid19 epoch, our control group. Subtracting this second difference from the first difference then provides a difference-in-difference estimate of the effect of the pandemic on the court outcome under scrutiny. In order for the labor courts in the pre-Covid19 epoch to serve as a good control group for the labor courts in the Covid19 epoch, in the absence of the pandemic the change in the court outcomes would have to be the same for both groups. Fig. 2 indicates that most court outcomes under consideration prior to March during the Covid19 and pre-Covid19 epochs indeed exhibit substantially parallel comovements, an indication that the parallel trends assumption would seem an apposite one to make in our context. In Section 5.2, we provide additional evidence in favor of the parallel trends assumption by using a dynamic specification and demonstrating the lack of pre-trends for the vast majority of the featured outcomes. Importantly, we have also verified that none of our findings are sensitive to an alternative definition of the pre-Covid 19 epoch.9 Nevertheless, the reader should keep in mind that the parallel trends assumption is inherently untestable. Our empirical strategy would be flawed if, for example, the labor courts during the pre-Covid19 epoch experienced unsuspected post-March shocks. In response to such concerns we emphasize that, based on our careful scrutiny of the Brazilian labor justice system during the time period under consideration, we did not discover any noteworthy events or developments that could invalidate our DD approach. In particular, from the start of the chosen pre-Covid19 epoch to the onset of the pandemic in 2020, the labor courts were not subject to any major institutional changes. Institutional history of the Brazilian labor justice system therefore lends credibility to our proposed methodology.Fig. 2 Temporal trends in court outcomes by epoch. Notes: The figure depicts the temporal evolution of the cross-sectional means, computed at the level of a regional labor court, of depicted variables for the Covid19 epoch (Nov 2019-Oct 2020) and the pre-Covid19 epoch (Nov 2018-Oct 2019). Fig. 2 The estimate of β based on specification (1) is informative of the average effect of the pandemic on the considered court outcomes. To gain further insight into whether, and if so how, the effect of the pandemic has varied over time, we estimate the following specification:(2) ycet=α+∑τβτMonthτ×Covide+γCovide+λt+μce+δSPJcet+εcet, where Monthτ is a dummy equal to 1 if the observation is from a specific month τ from the set of twelve months from November to the following October. As the month with respect to which we compare all month-by-month effects, we omit the month of February, that is, the month immediately preceding the start of the pandemic affecting the Covid19 epoch. The remaining elements of the right-hand-side of expression (2) are as defined under expression (1). Finally, we leverage the fact that, while the pandemic hit Brazil hard, not all regions were affected to the same extent and at the same time. We are therefore able to explore whether the effect of the pandemic on court outcomes has varied with the intensity of the pandemic as captured by the incidence of new Covid19 infections or Covid19-related deaths. To this end, we also examine the results based on the following model:(3) ycet=α+θPostt×Covide×Incidencecet + βPostt×Covide+γCovide+λt+μce+δSPJcet+εcet. In expression (3), Incidencecet is either the per-capita number of new Covid19 infections or the per-capita number of Covid19-related deaths in labor court region c during epoch e in month t. The remaining components of the right-hand-side of expression (3) coincide with those in expression (1). The primary coefficient of interest is θ, capturing the extent to which the effect of the pandemic on court outcome under investigation has varied with the intensity of the health crisis. We estimate all models using OLS. We base inference on heteroscedasticity-robust standard errors clustered at the level of the regional labor court. We therefore allow for correlation of unobservables within each regional labor court over time, an important feature given that we observe the same set of regional labor courts during both the Covid19 and the pre-Covid19 epoch. Because of a relatively small number of clusters (23), we for each coefficient estimate of interest also report the p-value for the test of the null of no effect, executed using wild bootstrap with 1000 replications (see Roodman et al., 2019). 5 Results and discussion 5.1 Average effect The results on the average effect of the pandemic are reported in Table 2 through 5. In each table, the odd-numbered columns report the results for the specification without any non-essential fixed effects or controls. The even-numbered columns show the results for the specification that includes court-in-epoch fixed effects and the staff per judge control. As anticipated given the exogenous nature of the pandemic, for each outcome, both specifications yield congruent estimates. In interpreting the results, we focus on the specifications featuring the full set of court-in-epoch fixed effects and the staff per judge control.Table 2 Effect on efficacy in adjudication. Table 2 Outcome: Clearance rate (adj.) Outcome: Pending per judge (adj.) (1) (2) (3) (4) Post March × Covid19 epoch −0.2243*** −0.2171*** 66.4289*** 62.3226*** (0.0432) (0.0457) (9.1613) (8.9006) [<0.000] [<0.000] [<0.000] [<0.000] Covid19 epoch FE Yes Yes Yes Yes Month FE Yes Yes Yes Yes Court-in-epoch FE No Yes No Yes Staff per judge control No Yes No Yes Observations 552 552 552 552 R-squared 0.3888 0.5779 0.0285 0.9503 Notes: The table presents OLS results based on the estimation of model (1). Post March is a dummy equal to one if observation is from the month of March or later in the applicable epoch. Covid19 epoch is a dummy equal to one if observation is from the Nov 2019-Oct 2020 period. Standard errors in parentheses are clustered at the level of regional labor court. In the brackets are pvalues for the test of the null of no effect executed using wild bootstrap with 1000 replications ***, **, and * denote statistical significance at the 0.1 %, 1%, and 5% level, respectively. Table 2 presents the results on the average effect of the pandemic on court efficacy in adjudication. The chief finding is apparent: the pandemic resulted in a quantitatively very large decrease in court efficacy in adjudication. Based on our estimates, the clearance rate in adjudication on average decreased by 0.22, an effect equal to about 20 percent of the average monthly clearance rate attained in the months from November to February in the Covid19 epoch or 18 percent of the average monthly clearance rate attained between March and October in the pre-Covid19 epoch. Notably, part (b) of Fig. 2 shows that the onset of the pandemic brought a decrease in the number of newly filed adjudication cases. The drop in the clearance rate in adjudication, implied by our estimates, is therefore not a consequence of an increase in demand for court services. Rather, the decrease in the courts' ability to meet the demand for adjudicatory services is driven by a fall in judicial output, as indicated in part (a) of Fig. 2. The large fall in the clearance rate in adjudication after March of the Covid19 epoch resulted in a post-March average monthly clearance rate notably smaller than one (see part A2 of Table 1), the critical value at which a court is just able to meet the incoming demand. Concurrently with the plummeting of the clearance rate, the number of pending adjudication cases per judge thus on average increased by as much as 62, an effect equal to about 26 percent of the average number of pending adjudication cases per judge in the months from November to February in the Covid19 epoch or 24 percent of the average number of pending adjudication cases per judge between March and October in the pre-Covid19 epoch. This is evidence that the pandemic resulted in a substantial increase in court backlogs. Table 3 shows the results on the average effect of the pandemic on the mode of disposition of adjudication cases. As one would have expected, we find a very sizeable effect of the pandemic on the share of cases resolved in trial hearings. Based on our estimates, the share of cases resolved in trial hearings on average decreased by 0.14, an effect equal to about 47 percent of the average monthly share of cases resolved in trial hearings in the months from November to February in the Covid19 epoch or 42 percent of the average monthly share of cases resolved in trial hearings between March and October in the pre-Covid19 epoch. We, however, do not find evidence that the pandemic, at least on average, affected the share of in-court settlements or the share of withdrawals, respectively.Table 3 Effect on modes of case disposition in adjudication. Table 3 Outcome: Share resolved in hearings Outcome: Share settled in-court Outcome: Share withdrawn (1) (2) (3) (4) (5) (6) Post March × Covid19 epoch −0.1365*** −0.1360*** −0.0164 −0.0169 −0.0037 −0.0042 (0.0270) (0.0282) (0.0098) (0.0107) (0.0044) (0.0046) [<0.000] [<0.000] [0.101] [0.108] [0.480] [0.425] Covid19 epoch FE Yes Yes Yes Yes Yes Yes Month FE Yes Yes Yes Yes Yes Yes Court-in-epoch FE No Yes No Yes No Yes Staff per judge control No Yes No Yes No Yes Observations 552 552 552 552 552 552 R-squared 0.3975 0.5980 0.2123 0.7132 0.0242 0.7832 Notes: See notes under Table 2. Table 4 displays the results on the average effect of the pandemic on judicial enforcement outcomes. In contrast to the profound detrimental effect of the pandemic on court efficacy in adjudication, we find no effect of the pandemic on court efficacy in enforcement. Neither the effect on the clearance rate in enforcement nor the effect on the volume of pending enforcement cases per judge are statistically significantly different from zero. These findings may be explained by three key features of the Brazilian labor-court enforcement proceedings. First, enforcement proceedings do not require the conduct of hearings, rendering enforcement comparatively less vulnerable to the pandemic-caused disruption. Second, labor-court judges are able to implement many enforcement-related tasks, such as the confiscation of funds from the checking account of a non-compliant losing party or the blocking of the transfer of their property title, using remotely-accessible software (see Castelliano et al., 2020b). Third, it is likely that, following the onset of the pandemic, the judicial and staff effort that would have normally been devoted to the conduct of trial hearings and other adjudication activities was purposefully re-directed toward enforcement activities.Table 4 Effect on judicial enforcement. Table 4 Outcome: Clearance rate (enf.) Outcome: Pending per judge (enf.) Outcome: Share enforced payments (1) (2) (3) (4) (5) (6) Post March × Covid19 epoch 0.0495 0.0573 −3.4881 −11.1254 0.0256 0.0229 (0.1336) (0.1385) (11.8367) (10.9191) (0.0198) (0.0207) [0.706] [0.672] [0.776] [0.288] [0.223] [0.286] Covid19 epoch FE Yes Yes Yes Yes Yes Yes Month FE Yes Yes Yes Yes Yes Yes Court-in-epoch FE No Yes No Yes No Yes Staff per judge control No Yes No Yes No Yes Observations 552 552 552 552 552 552 R-squared 0.1660 0.5355 0.0008 0.9747 0.0269 0.5955 Notes: See notes under Table 2. Notably, we also find no evidence of an effect of the pandemic on the share of payments executed upon judicial enforcement. In particular, the pandemic and the accompanying economic downturn have, at least on average, evidently not resulted in an increased use of judicial enforcement as means to securing the execution of court-sanctioned payments stemming from the resolution of labor disputes. Finally, Table 5 shows the results for the effect of the pandemic on the composition of new filings with respect to enforcement and adjudication matters. The estimates reveal that, as a result of the pandemic, the share of new filings requiring judicial enforcement on average increased by 0.06, an effect equal to about 18 percent of the mean monthly share of newly filed enforcement cases between November to February in the Covid19 epoch or between March and October in the pre-Covid19 epoch. This is evidence that, even though the judicial enforcement outcomes do not seem to have been impacted, the pandemic has altered the composition of the demand for labor court services. As a consequence of the pandemic, there has been an increase in the demand for enforcement relative to adjudication. As parts (b) and (i) of Fig. 2 indicate, much of this composition effect can be attributed to a reduction in the overall demand for adjudication.Table 5 Effect on composition of new filings. Table 5 Outcome: Share enforcement new filings (1) (2) Post March × Covid19 epoch 0.0589*** 0.0581*** (0.0115) (0.0122) [<0.000] [<0.000] Covid19 epoch FE Yes Yes Month FE Yes Yes Court-in-epoch FE No Yes Staff per judge control No Yes Observations 552 552 R-squared 0.2178 0.6604 Notes: See notes under Table 2. 5.2 Month-by-month effect The results in Tables 2 through 5 show the average effect of the pandemic, but do not reveal whether, and if so how, the effect perhaps varied from month to month. Yet we know that the pace of the pandemic itself has varied over time and, moreover, that, after the pandemic's onset, the Brazilian labor courts adopted different operational responses at different points in time (see Fig. 1). To capture the resulting dynamics, we turn to investigating the results based on specification (2). We summarize our findings with the aid of Fig. 3 . Each part of Fig. 3 summarizes the results for a specific outcome variable, showing the month-by-month effects from November 2019 to October 2020. The omitted (comparison) effect is that for February, the month immediately preceding the onset of the pandemic. For each month, we display the point estimate and the corresponding 95-percent confidence interval computed using heteroscedasticity-robust standard errors clustered at the level of the regional labor court.Fig. 3 Summary of month-by-month effects. Notes: The figure shows the point estimates and the corresponding 95-percent confidence intervals based on OLS estimates of model (2) for different outcome variables. The omitted comparison month effect is that for February Fig. 3 Parts (a) and (b) of Fig. 3 show the month-by-month effect of the pandemic on court efficacy in adjudication. Part (a) illustrates that the adverse effect on the courts' capacity to meet the ongoing demand for adjudication was especially large, and in fact increased in terms of the absolute magnitude, over the initial months of the pandemic. The deleterious effect was strongest in the month of May, when the decrease in the clearance rate exceeded 0.4, an effect equal to about 40 percent of the average monthly clearance rate attained in the months from November to February in the Covid19 epoch or 35 percent of the average monthly clearance rate attained between March and October in the pre-Covid19 epoch. After May, as the courts gradually managed to regroup and introduced virtual hearings (see Section 2.2), the negative effect on the clearance rate weakened, even if it did not fully disappear. This is evidence that, after about half a year after the start of the pandemic, the Brazilian labor courts did eventually find a way to at least partly cope with the demand for adjudication. Part (b) of Fig. 3 traces out the month-by-month effect on the stock of pending adjudication cases per judge. Given the persistent and large negative effect on the clearance rate, which pushed the monthly clearance rate below the benchmark value of one, the pandemic resulted in persistent accumulation of unresolved cases. The effect on the increase in backlogs is largest for the final month covered by our data (October 2020), when the increase in the stock of pending adjudication cases per judge relative to February of the same year was equal to about 111 cases. This is an effect of the size of about 47 percent of the volume of pending adjudication cases per judge during the months from November to February in the Covid19 epoch or 43 percent of the volume of pending adjudication cases per judge during the months from March to October in the pre-Covid19 epoch. The documented effect on court backlogs is an indication that the pandemic will likely have a lasting detrimental impact on the ability of the Brazilian labor courts to administer justice in a timely manner. Parts (c) through (e) of Fig. 3 show the month-by-month effect of the pandemic on the modes of disposition of adjudication cases. Predictably, since its onset, the month-by-month effect of the pandemic has been a reduction in the share of cases resolved via trial hearings. The reduction was most pronounced in the months of April and May, when the share of cases resolved via hearings dropped by around 0.3, bringing the average share of cases resolved via trial hearings effectively to zero (the average monthly share of cases resolved via trial hearings was about 0.30 between November and February in the Covid19 epoch and about 0.33 between March and October in the pre-Covid19 epoch). Furthermore, the negative effect on the share of cases resolved via trial hearings has been a persistent one. Thus, the transition of courts to virtual hearings in the late spring alleviated, but did not eliminate, the deleterious effect of the pandemic on the ability of the courts to rely on hearings to resolve disputes. Based on the estimates in columns (4) through (6) of Table 3, the average effect of the pandemic on the share of cases disposed via settlement has been indistinguishable from zero. Part (d) of Fig. 3, however, shows evidence of intriguing dynamics with regard to month-by-month effect on the share of in-court settlements. Right after the onset of the pandemic, the effect was negative and very large: in April, for example, the pandemic resulted in a decrease in the share of adjudication cases settled in-court of about 0.17, an effect equal to approximately 48 percent of the mean monthly value of the share of cases settled in-court between November and February in the Covid19 epoch or 46 percent of the same value between March and October in the pre-Covid19 epoch. This pattern is consistent with the interpretation that the initial suspension of hearings and nation-wide restrictions on in-person interaction obstructed the exchange of information between the disputing parties, a process that normally facilitates settlement via the convergence of the disputing parties' views about the likely trial outcome (Boyd and Hoffman, 2013; Bielen et al., 2017, 2020). From June, however, the month-by-month effect on the share of cases settled in-court becomes positive, with the magnitude of the effect reaching about 18 percent of the average monthly share of cases settled in-court between November and February in the Covid19 epoch or between March and October in the pre-Covid19 epoch. That is, once the courts began to implement virtual hearing sessions, the pandemic's effect on the share of in-court settlements turned to positive. It appears, therefore, that the combination of the re-instituted possibility for court-facilitated exchange of information among the disputing parties and the judges, the awareness about the mounting court backlogs with corresponding increased prospects of court delays, and the unavoidable economic turmoil increased the relative attractiveness of settlement as means to resolution of labor disputes. On the other hand, congruent with the estimates of the average effect of the pandemic on the share of withdrawals, noted in the previous section, we find no evidence of an effect on this particular court outcome during any of the months. The pandemic has thus not noticeably altered the plaintiffs' incentives to altogether abandon their claims. The remaining parts of Fig. 3 summarize the results for the month-by-month effect of the pandemic on the enforcement outcomes (parts (f) through (h)) and the composition of incoming cases with respect to enforcement versus adjudication (part (i)). In line with the estimates of the average effect, reported in Table 4, we see little evidence of an effect of the pandemic on the efficacy of the enforcement aspects of labor court operations (parts (f) and (g)). In the month of April, there was a temporary positive effect on the clearance rate in enforcement. This rather peculiar effect, however, is driven primarily by a temporary drop in the number of newly filed enforcement cases in that month (see part (i) of Fig. 2), an artifact of the administrative chaos that accompanied the onset of the pandemic and impacted case registration processes across all labor courts. The month-by-month results also indicate that the pandemic temporarily and intermittently increased the reliance of litigants on judicial enforcement as means of executing court-ordered payments. The effect of the pandemic on the share of court-endorsed payments secured via judicial enforcement is positive and statistically significant (at five-percent level) in the months of June and August. The impact, however, dissipates by September (part (h)). In contrast, congruent with the results reported in Table 5, the impact of the pandemic on the composition of newly filed labor-court cases with respect to enforcement versus adjudication has been comparatively more long-lived (part (i)). Reflecting a drop in the overall demand for adjudication (see part (b) of Fig. 2), the pandemic resulted in a sizeable and enduring increase in the relative demand for enforcement versus adjudication. Last but not least, Fig. 3 illustrates that seven out of the nine court outcomes under consideration do not exhibit any unwanted pre-trends in the months prior to March. The share of cases resolved via hearings (part (c)) and the share of in-court settlements (part (d)) show some limited evidence of pre-trends. For the corresponding outcomes, our point estimates should thus be interpreted with some caution. All in all, however, the insights based on Fig. 3 complement the insights based on Fig. 2 in lending support for the use of the difference-in-difference approach in our setting. 5.3 Effect heterogeneity by pandemic intensity Tables 6 through 9 present the results on the heterogeneity of the effect of the pandemic by the intensity of the pandemic. In each of the tables, odd-numbered columns show the results when we measure the intensity of the pandemic using the rate of new Covid19 cases per 1000 people. Even-numbered columns display the results when we measure the intensity of the pandemic using the rate of Covid19-related deaths per 10,000 people.Table 6 Effect heterogeneity by pandemic intensity, efficacy in adjudication. Table 6 Outcome: Clearance rate (adj.) Outcome: Pending per judge (adj.) (1) (2) (3) (4) Post March × Covid19 epoch × New Covid19 cases (per 1 K) −0.0154 3.6297*** (0.0079) (0.8226) [0.045] [<0.000] Post March × Covid19 epoch × Covid19 deaths (per 10 K) −0.1053** 12.1539*** (0.0276) (2.5555) [0.001] [<0.000] Post March × Covid19 epoch −0.1670** −0.1290* 49.2408*** 51.4456*** (0.0506) (0.0454) (7.8229) (7.8418) [0.005] [0.009] [<0.000] [<0.000] Covid19 epoch FE Yes Yes Yes Yes Month FE Yes Yes Yes Yes Court-in-epoch FE Yes Yes Yes Yes Staff per judge control Yes Yes Yes Yes Observations 552 552 552 552 R-squared 0.4891 0.5159 0.9540 0.9528 Notes: The table presents OLS results based on the estimation of model (3). Post March is a dummy equal to one if observation is from the month of March or later in the applicable epoch. Covid19 epoch is a dummy equal to one if observation is from the Nov 2019-Oct 2020 period. New Covid19 cases and Covid19 deaths are measures of pandemic intensity (see Table A1 in the Appendix) that vary both across labor court regions and over time. Standard errors in parentheses are clustered at the level of regional labor court. In the brackets are p-values for the test of the null of no effect executed using wild bootstrap with 1000 replications. ***, **, and * denote statistical significance at the 0.1 %, 1%, and 5% level, respectively. We find, first and foremost, that the intensity of the pandemic has been an important moderating factor with regard to several court outcomes. Based on the results in Table 6, the detrimental effect of the pandemic on the ability of the courts to meet the ongoing demand for adjudication and on the stock of adjudicatory backlogs has been larger in the region-months characterized by greater pandemic intensity. For example, based on the estimates in columns (2) and (4) of Table 6, relative to the baseline effect of the pandemic, a one-standard-deviation increase in the number of Covid19-related deaths per 10,000 inhabitants is associated with an additional decrease in the clearance rate of 0.36 and an additional increase of 41 pending adjudication cases per judge. Thus, the severity of the health crisis has importantly shaped the effect of the pandemic on court efficacy in adjudication by exacerbating the pandemic's adverse impact. In addition, the intensity of the pandemic has played a role in shaping the effect of the pandemic on the court modes of case disposition in adjudication, and in particular on the use of in-court settlements (see Table 7 ). The effect of the pandemic on the share of in-court settlements is negative for the region-months that exhibit low pandemic intensity, but positive for the region-months that exhibit high pandemic intensity. Based on the estimates in columns (3) and (4) of Table 7, the effect of the pandemic on the share of in-court settlements turns from negative to positive as the number of new Covid19 cases per 1000 inhabitants reaches about 4.5 (about 117 percent of the mean monthly number of new Covid19 cases per 1000 inhabitants since March 2020) or, alternatively, the number of Covid19-related deaths per 10,000 inhabitants reaches about 1.2 (about 123 percent of the mean monthly number of Covid19-related deaths per 10,000 inhabitants since March 2020).Table 7 Effect heterogeneity by pandemic intensity, modes of disposition in adjudication. Table 7 Outcome: Share resolved in hearings Outcome: Share settled in-court Outcome: Share withdrawn (1) (2) (3) (4) (5) (6) Post March × Covid19 epoch × New Covid19 cases (per 1 K) −0.0002 0.0131*** −0.0002 (0.0025) (0.0023) (0.0003) [0.939] [<0.000] [0.380] Post March × Covid19 epoch × Covid19 deaths (per 10 K) −0.0184 0.0449*** −0.0008 (0.0128) (0.0097) (0.0016) [0.185] [0.001] [0.627] Post March × Covid19 epoch −0.1327*** −0.1172*** −0.0589*** −0.0519*** −0.0036 −0.0036 (0.0221) (0.0222) (0.0091) (0.0096) (0.0046) (0.0040) [<0.000] [<0.000] [<0.000] [<0.000] [0.431] [0.354] Covid19 epoch FE Yes Yes Yes Yes Yes Yes Month FE Yes Yes Yes Yes Yes Yes Court-in-epoch FE Yes Yes Yes Yes Yes Yes Staff per judge control Yes Yes Yes Yes Yes Yes Observations 552 552 552 552 552 552 R-squared 0.4458 0.4521 0.5987 0.5786 0.7641 0.7641 Notes: See notes under Table 6. Interestingly, the intensity of the pandemic appears to have played no role in moderating the effect of the pandemic on the share of cases resolved via trial hearings (columns (1) and (2) of Table 7). This finding is likely a reflection of the fact that, upon the onset of the pandemic, trial hearings came to a complete halt across all Brazilian labor courts at roughly the same time. Trial hearings eventually also resumed, albeit in a restricted (virtual) format, across all labor courts after the month of May (see Section 2.2). In this sense, the intensity of the pandemic per se has therefore not been a salient moderating factor. We also find no evidence of the importance of the intensity of the pandemic as a moderator of the effect of the pandemic on any of the judicial enforcement outcomes (see Table 8 ). More generally, congruent with the findings reported in Section 5.1, the effect of the pandemic on any of the judicial enforcement outcomes remains undetectable even upon allowing the effect of the pandemic to vary with pandemic intensity.Table 8 Effect heterogeneity by pandemic intensity, judicial enforcement. Table 8 Outcome: Clearance rate (enf.) Outcome: Pending per judge (enf.) Outcome: Share enforced payments (1) (2) (3) (4) (5) (6) Post March × Covid19 epoch × New Covid19 cases (per 1 K) −0.0133 −1.7954 0.0048 (0.0089) (2.9124) (0.0027) [0.114] [0.847] [0.060] Post March × Covid19 epoch × Covid19 deaths (per 10 K) −0.0435 −5.9170 0.0079 (0.0341) (9.7524) (0.0075) [0.193] [0.917] [0.274] Post March × Covid19 epoch 0.1139 0.1047 −5.0054 −6.1796 0.0079 0.0179 (0.1370) (0.1396) (8.5330) (8.0094) (0.0233) (0.0226) [0.407] [0.449] [0.550] [0.441] [0.731] [0.421] Covid19 epoch FE Yes Yes Yes Yes Yes Yes Month FE Yes Yes Yes Yes Yes Yes Court-in-epoch FE Yes Yes Yes Yes Yes Yes Staff per judge control Yes Yes Yes Yes Yes Yes Observations 552 552 552 552 552 552 R-squared 0.4909 0.4901 0.9748 0.9747 0.5742 0.5704 Notes: See notes under Table 6. The intensity of the pandemic, however, has shaped the effect of the pandemic on the composition of new filings. According to the estimates in Table 9 , a one-standard-deviation increase in the number of Covid19-related deaths per 10,000 inhabitants, or, alternatively, an equivalent increase in the number of new Covid19 infections per 1000 inhabitants, is associated with an additional increase in the share of new filings that necessitate judicial enforcement of about 0.06. Hence, all else equal, the relative demand for judicial enforcement has been greater in areas and time periods exhibiting greater pandemic intensity.Table 9 Effect heterogeneity by pandemic intensity, composition of new filings. Table 9 Outcome: Share enforcement new filings (1) (2) Post March × Covid19 epoch × New Covid19 cases (per 1 K) 0.0090*** (0.0018) [<0.000] Post March × Covid19 epoch × Covid19 deaths (per 10 K) 0.0346*** (0.0068) [<0.000] Post March × Covid19 epoch 0.0273* 0.0287* (0.0103) (0.0123) [0.011] [0.020] Covid19 epoch FE Yes Yes Month FE Yes Yes Court-in-epoch FE Yes Yes Staff per judge control Yes Yes Observations 552 552 R-squared 0.6259 0.6229 Notes: See notes under Table 6. 6 Summary and conclusion We have provided the first systematic empirical analysis of the consequences of the Covid19 pandemic for the performance of courts at accomplishing their primary function, the disposition of cases. Using a newly-assembled, monthly panel of Brazilian regional labor courts and employing a difference-in-difference approach, we have demonstrated, first, that the pandemic has had a very large and persistent deleterious impact on adjudicatory efficacy. Timing-wise, the adverse effects on adjudicatory efficacy were especially drastic in the first few months after the onset of the pandemic, when many court activities were altogether suspended. Upon the eventual introduction of virtual court hearings and other operational measures intended to facilitate adjudication, the drop in the clearance rate of adjudication cases decreased somewhat in absolute magnitude, but, at least by the end of our observation window, never vanished. More generally, the adverse effect on court efficacy has been largest in region-months where the rates of new Covid19 infections or Covid19-related deaths have been highest. Second, the pandemic has affected the modes through which the labor courts resolve disputes. As anticipated, following the suspension of in-person activities, including trial hearings, the share of cases resolved in hearings effectively dropped to zero. Concurrently, the share of cases resolved via in-court settlement fell as well. This is consistent with the interpretation that the restraints on in-person interaction and suspension of hearings limited the inter-party exchange of information about the cases, which in turn reduced the prospects of settlement. However, once the courts instituted virtual hearing sessions, as the prospects of court delays became apparent, and when a prolonged recession was clearly in sight, the effect of the pandemic on the share of in-court settlements turned to positive. Resonating with this explanation, we also find that the effect of the pandemic on the share of in-court settlements was positive in region-months characterized by the highest rate of new Covid19 infections or the largest Covid19-related death toll rate. The pandemic, however, did not affect the share of cases disposed as a result of withdrawals of already started lawsuit. Third, we find little evidence of an effect of the pandemic on judicial enforcement outcomes. In particular, our estimates reveal that the pandemic has not affected court efficacy in the context of judicial enforcement, even though the pandemic has increased the relative demand for enforcement versus adjudication. Because judicial enforcement of final decisions in Brazilian labor courts can be effort-intensive and, indeed, constitutes a sizeable portion of the courts' dockets, our findings indicate that the pandemic-induced transition of the Brazilian labor court judges toward a remote completion of the enforcement tasks was overall quite effective. We also find only limited evidence that the pandemic has increased the need for judicial enforcement as means of securing the execution of court-sanctioned payments. To the extent than such an effect is detectable, the effect has been intermittent and transitory. During the first eight months from the onset of the pandemic, we therefore do not see much indication that the losing parties (e.g. employers), who are mandated by the court to compensate the winning parties (e.g. employees), have, perhaps as a reaction to the economic recession, strategically chosen to avoid the execution of court-ordered payments. This particular result is perhaps the most uplifting of all findings that have emerged from our analysis. It suggests that, at least when it comes to the timing of received compensation, awarded on the basis of unlawful violation of labor contracts, the employees who are eligible for such compensation and who often stem from the most vulnerable segments of the Brazilian society have, on average, not really been adversely impacted by the pandemic. Overall, however, one may anticipate that the pandemic will continue to exert a lasting negative impact on the ability of the Brazilian labor courts to deliver justice, thereby exacerbating already considerable pre-existing socio-economic inequality. Our goal in this paper has been to offer an empirical analysis of the effect of the Covid19 pandemic on some of the most salient aspects of court operations. We have done so in the context of Brazilian labor justice, utilizing rich up-to-date court-level data. Future research will undoubtedly uncover many more aspects in which the pandemic has already impacted and will in the foreseeable future continue to impact the functioning of courts and the administration of justice both in Brazil and in other jurisdictions worldwide. Especially in the context of labor justice it would be pertinent to examine whether, and if so in what way, the pandemic has affected the composition of court dockets with respect to case complexity and the exact substance of labor-related claims. The resulting analysis would also help illuminate the extent to which the documented efficacy effects of the pandemic can be attributed to the operational responses of the courts versus any case composition effects. Investigation of such questions will require access to case-level data. At the same time, when even more recent court-level data become available, it will be important to provide updated estimates of the consequences of the pandemic for the outcomes explored in the present paper. Indeed, at the time of our writing, Brazil is already amidst a new devastating wave of the spread of the virus. In the current time, when the pandemic is nowhere close to under control, medical and natural-science research aimed at preventing, to the greatest extent possible, further loss of human life and securing adequate healthcare responses to the ravaging virus should remain the highest priority. For social scientists and other scholars, however, it will be important to enhance our understanding of the impact of the pandemic on key societal institutions, including the courts of law, and propose appropriate policy responses to help mitigate the associated rapidly rising social costs. Such research will necessarily require evidence-based insight of the type that we have striven to provide in the current paper. Declaration of Competing Interest None. Appendix Table A1 Variable definitions. Table A1Variable Definition Courts, adjudication  Clearance rate (adj.) The number of resolved adjudication cases during a month divided by the number of newly filed adjudication cases during the same month.  Pending per judge (adj.) The number of adjudication cases pending at the end of a month divided by the number of serving judges during the same month.  Share resolved in hearings The number of adjudication cases resolved in trial hearings during a month divided by the total number of resolved adjudication cases during the same month.  Share settled in-court The number of adjudication cases settled in-court during a month divided by the total number of resolved adjudication cases during the same month.  Share withdrawn The number of withdrawn adjudication cases during a month divided by the total number of resolved adjudication cases during the same month. Courts, enforcement  Clearance rate (enf.) The number of completed enforcement proceedings during a month divided by the number of newly initiated enforcement cases during the same month.  Pending per judge (enf.) The number of enforcement cases pending at the end of the month divided the number of serving judges during the same month.  Share enforced payments The value of payments executed upon completed judicial enforcement proceedings during a month divided by the total value of all executed payments during the same month. Courts, new filings composition  Share enforcement new filings The number of newly initiated enforcement cases during a month divided by the number of all new (adjudication and enforcement) case filing during the same month. Courts, staffing  Staff per judge The number of judicial support staff (judicial assistants and administrative staff) during a month divided by the number of serving judges during the same month. Pandemic intensity  New Covid19 cases (per 1,000 people) New Covid19 cases per capita in a given month, multiplied by 1,000.  Covid19 deaths (per 10,000 people) Covid19-related deaths per capita in a given month, multiplied by 10,000. Notes: The table provides the definitions of the outcome and select other variables used to generate the estimates shown in Table 1, Table 2, Table 3, Table 4, Table 5, Table 6, Table 7 and Fig. 3. Acknowledgement For helpful comments and suggestions, we thank an anonymous reviewer and Eric Helland, our editor. 1 See, for example, the RAND Corporation's "COVID19 and the Courts" virtual event (https://www.rand.org/events/2020/10/01.html). 2 See https://coronavirus.jhu.edu/data/mortality. 3 The information presented in this section draws heavily on parts of Section 2 in Castelliano et al. (2020b). 4 Because each labor court office consists of a titled and a substitute judgeship and since the allocation of cases between the two judgeship types within each office is random, the enforcement proceedings are not necessarily carried out by the same judge that adjudicated the original case. 5 Access to justice was further facilitated by the fact that many states permitted the re-opening of law offices soon after the initial March lockdown. In addition, many law offices swiftly transitioned to offering their services online. 6 The labor court regions normally coincide with Brazilian state borders. The exceptions are the 8th, 10th, 11th, and 14th labor court region, each of which extends over two states. In addition, the state of São Paulo comprises two labor court regions: the 2nd region (the city of São Paulo) and the 15th region (the remaining part of the state). 7 At the time of our conducting of this research and the writing up of the results, more recent monthly labor-court data are not (yet) available. 8 Because we view the state of São Paulo as a single labor court region and because the remaining labor court regions either coincide with state borders or precisely absorb two states (see note 6 above), the merger of the regional labor court data and Covid19 data is exact. 9 As a robustness check, we redefined the pre-Covid19 epoch as the November 2016-October 2017 period, the latest suitable alternative 12-month span between November and the following October for which monthly labor court data are available to us. (The November 2017-October 2018 period is not an appropriate choice for the purpose at hand because the end of year 2017 encompasses the start and implementation of a major reform.) The results obtained on the basis of this alternative definition of the pre-Covid19 epoch were both qualitatively and quantitatively very similar to the results reported in Section 5. ==== Refs References Baldwin Julie Marie Eassey John M. Brooke Erika J. Court operations during the COVID-19 pandemic Am. J. Crim. Justice 2020 10.1007/s12103-020-09553-1 forthcoming Beenstock Michael Haitovsky Yoel Does the appointment of judges increase the output of the judiciary? Int. Rev. Law Econ. 24 3 2004 351 369 Bełdowski Jarosław Dąbroś Łukasz Wojciechowski Wiktor Judges and court performance: a case study of district commercial courts in Poland Eur. J. Law Econ. 50 1 2020 171 201 Bielen Samantha Grajzl Peter Marneffe Wim Procedural events, judge characteristics, and the timing of settlement Int. Rev. Law Econ. 52 2017 97 110 Bielen Samantha Grajzl Peter Marneffe Wim The resolution process and the timing of settlement of medical malpractice claims Health Econ. Policy Law 15 4 2020 509 529 30994084 Boyd Christina L. Hoffman David A. Litigating toward settlement J. Law Econ. Organ. 29 4 2013 898 929 Buscaglia Edgardo Ulen Thomas S. A quantitative assessment of the efficiency of the judicial sector in Latin America Int. Rev. Law Econ. 17 2 1997 275 291 Castelliano Caio Grajzl Peter Alves Andre Watanabe Eduardo Adjudication forums, specialization, and case duration: evidence from brazilian federal courts Justice Syst. J. 2020 10.1080/0098261X.2020.1854905 forthcoming Castelliano Caio Grajzl Peter Guimaraes Tomas A. Alves Andre "Judicial enforcement and caseload heterogeneity: theory and evidence from Brazil Working Paper, Presented at the EMLE Midterm Meeting University of Hamburg, Germany 2020 CEPEJ - The European Commission for the Efficiency of Justice National Judiciaries’ COVID-19 Emergency Measures of COE Member States 2020 https://www.coe.int/en/web/cepej/national-judiciaries-covid-19-emergency-measures-of-coe-member-states CEPEJ - The European Commission for the Efficiency of Justice European Judicial Systems–Edition 2020 (2018 data): Efficiency and Quality of Justice 2020 Council of Europe Strasbourg, France Chemin Matthieu. The impact of the judiciary on entrepreneurship: evaluation of Pakistan’s’ Access to justice programme’ J. Public Econ. 93 1–2 2009 114 125 Christensen Robert K. Szmer John Examining the efficiency of the U.S. courts of appeals: pathologies and prescriptions Int. Rev. Law Econ. 32 1 2012 30 37 CNJ - Conselho Nacional de Justiça Relatório Justiça Em Números 2020: Ano-base 2019 2020 Brasil: Conselho Nacional de Justiça Brasilia Di Vita Giuseppe. Factors determining the duration of legal disputes: an empirical analysis with Micro data J. Inst. Theor. Econ. 168 4 2012 563 587 Dimitrova-Grajzl Valentina Grajzl Peter Sustersic Janez Zajc Katarina Court output, judicial staffing, and the demand for court services: evidence from slovenian courts of first instance Int. Rev. Law Econ. 32 1 2012 19 29 Dimitrova-Grajzl Valentina Grajzl Peter Zajc Katarina Understanding modes of civil case disposition: evidence from slovenian courts J. Comp. Econ. 42 4 2014 924 939 EBRD - European Bank for Reconstruction and Development Law in Transition 2014 2014 European Bank for Reconstruction and Development. London, UK Engstrom David Freeman Post-COVID courts UCLA Law Review Discourse 68 2020 246 267 Galanter Marc. The vanishing trial: an examination of trials and related matters in federal and state courts J. Empir. Leg. Stud. 1 3 2004 459 570 Grajzl Peter Silwal Shikha Multi-court judging and judicial productivity in a career judiciary: evidence from Nepal Int. Rev. Law Econ. 61 2020 105888 Marciano Alain Melcarne Alessandro Ramello Giovanni B. The economic importance of judicial institutions, their performance and the proper way to measure them J. Inst. Econ. 15 1 2019 81 98 Matyas David Wills Peter Dewitt Barry Imagining Resilient Courts: From COVID to The Future of Canada’s Judicial System SSRN Working Paper No.3778869 2021 McIntyre Joe Olijnyk Anna Pender Kieran Civil courts and COVID-19: challenges and opportunities in Australia Altern. Law J. 45 3 2020 195 201 Pistor Katharina Law in the Time of COVID-19 2020 Columbia Law School. New York, NY Puddister Kate Small Tamara A. Trial by zoom? The response to COVID-19 by Canada’s courts Can. J. Political Sci. 53 2020 373 377 Roodman David Nielsen MortenØrregaard MacKinnon James G. Webb Matthew D. Fast and wild: bootstrap inference in Stata Using Boottest Stata J. 19 2019 4 60 Rosales-López Virginia Economics of court performance: an empirical analysis Eur. J. Law Econ. 25 3 2008 231 251 Sourdin Tania Zeleznikow John Courts, mediation and COVID-19 Australian Bus. Law Review 2020 forthcoming TST - Tribunal Superior do Trabalho Relatório Geral Da Justiça Do Trabalho 2020 Brasil: Secretaria-Geral da Presidência do TST Brasilia Voigt Stefan. Determinants of judicial efficiency: a survey Eur. J. Law Econ. 42 2 2016 183 208 Warner Randall H. Judging in a time of COVID Fam. Court Rev. 58 4 2020 965 967

0

PMC9754154

NO-CC CODE

2022-12-16 23:26:28

no

Radiother Oncol. 2022 May 31; 170:S856

latin-1

Radiother Oncol

10.1016/S0167-8140(22)04097-X

oa_other

==== Front Landsc Urban Plan Landsc Urban Plan Landscape and Urban Planning 0169-2046 0169-2046 Published by Elsevier B.V. S0169-2046(21)00205-X 10.1016/j.landurbplan.2021.104242 104242 Perspective Essay A landscape planning agenda for global health security: Learning from the history of HIV/AIDS and pandemic influenza Spencer James Nguyen H. Louisiana State University, West David Boyd Hall, Baton Rouge, LA 70803, USA 15 9 2021 12 2021 15 9 2021 216 104242104242 19 11 2020 25 8 2021 2 9 2021 © 2021 Published by Elsevier B.V. 2021 Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. This paper considers the role of landscape planning and design in the context of a growing need for research and policy recommendations associated with Emerging Infectious Diseases (EIDs), of which COVID-19 is the most recent. Beginning with a definition of EIDs and their origins within the context of landscape planning, the paper then argues that planning and design scholars and practitioners should begin by seeing the importance of a “global urban ecosystem” (GUE) comprised of rapidly transforming metropolitan and regional “patches” connected through “corridors” of relatively unregulated global transportation and mobility networks. It then revisits the history of the two prior global pandemics of HIV/AIDS and pandemic influenza to establish the importance of a landscape planning perspective at the intersection of wildlife, livestock, and globally connected human communities. The essay concludes by arguing that this GUE concept can facilitate creative planning and design by adapting concepts established in other patch and corridor networks like urban transit systems to the ongoing risk of future pandemic EIDs. Keywords Emerging Infectious Disease HIV/AIDs Influenza Global urban ecosystem COVID19 Pandemic ==== Body pmc1 Introduction Where did the COVID-19 virus come from? More specifically, how did this regional epidemic become a global pandemic? In this paper I use the conventional distinction that an “epidemic” is an unexpected rise in cases of any disease within a community, while a “pandemic” is an epidemic that crosses national and/or global borders, generally creating new epidemics in distant places. It is essential to answer origin questions about pandemics – and not just epidemics – in ways that lead towards policies for prevention and management of future pandemics. Since COVID-19 is the latest and most widespread of many recent zoonotic diseases, it is important to ask about its origin not simply to clarify responsibility for the current problems, but as a more pro-active starting point for understanding what can be done to mitigate the risk of future pandemics. In this paper, I argue that landscape and urban planning should be an area of study at the forefront of answering the question, and outline a theoretical perspective on why the biotic and built landscapes, as well as the human behaviors happening within them are so important. Moreover, I assert that this landscape and urban planning perspective must be complemented by an understanding of how distant landscapes are connected through global networks of landscape patches. Future pandemic mitigation strategies need to go well beyond the current range of policy alternatives that have emphasized medical and biological interventions, and this theoretical framework I describe below can help guide these strategies. Even before the advent of COVID-19, public health agencies worldwide had become very concerned with how infectious disease outbreaks take hold in the world’s most vulnerable areas – often times in countries with few health and other resources needed to stem the tide of infection locally before it becomes a global threat (CDC, 2017). Pandemics have become a growing security concern for many countries, and even three years prior to COVID-19, the CDC wrote that the United States’ national security was at risk within 36 h of a pathogen outbreak in any remote area of the world (CDC, 2017). As COVID-19 has shown very clearly, compounding the dire health threats are economic impacts imposed by potential slowdowns of global trade and movements enabled by today’s extensive global connectivity of people and goods. With large numbers of passengers traveling by air each year, and many of them crossing national borders, EID’s have long posed a significant health and economic threat that came to light in late 2019 (CDC, 2020b). This planetary mobility of people has been mirrored by an equally expansive growth in trade and container shipping of goods (Cudahy, 2006), which also should push us to think about the origins of pandemics clearly and thoughtfully. This global impact points towards mitigation strategies based on what I call (Spencer, 2014) the global urban ecosystem (GUE) comprised of dense mixing-bowl-like patches of physical, tangible, built and biotic landscapes managed by local communities, and connected to one another by planetary corridors of travel and transportation. The importance of this network of landscape patches and corridors of the GUE should be underlain by an understanding of the category of diseases that become pandemics. Emerging Infectious Diseases (EIDs) comprise a category of illnesses that pose the greatest threat of pandemics because – by definition – so little is understood about identifying, treating, and preventing them. 2 Emerging Infectious Diseases: what are they and what is known about them? A growing literature describes the outbreaks of EIDs in remote areas of the globe as the result of several convergent conditions: 1) infectious zoonotic pathogens “spilling over” from animals to humans; 2) unplanned and rapid urbanization; 3) agricultural intensification; 4) the development of antimicrobial resistance; and 5) weak public health infrastructures (Wilcox and Colwell, 2005, Kapan et al., 2006, Neiderud, 2015, Boyce et al., 2019, Wu et al., 2017, Daszak et al., 2009). The first three of these five drivers of EIDs center on landscape-level trends. Such studies argue and document that intensive farming practices and dense urban populations create perfect “mixing bowl” environments for the evolution of new human viruses. Others have also detailed and summarized for planners the specific hypothesized mechanisms for virus evolution into new pathogens, and proposed some preliminary thoughts for controlling them at their source at low financial and political cost (Spencer, Marasco, & Eichinger, 2021). This practical planning orientation aligns with current developments in public health practice. Recently, public health practitioners have highlighted the importance of understanding the landscape patches where human and animals interact intimately, extensively, and regularly. The “one health” approach promoted by the CDC and supported by over 20 national governments across the developing and developed world (e.g. CDC, 2020a, CDC, 2020b) recognizes that a shared physical environment between humans and animals means that the protection of human health depends on the protection of animal health. According to these professionals, regular animal contact with humans and with one another necessitates that we understand the health status of both wild and domesticated animals, their relationships with one another, and with the humans they live next to. In these mixing bowls, pathogens move seamlessly across species boundaries because landscapes of development and rapid change are where humans and animals come into dense proximity without mitigating plans and designs. Thus, identifying landscapes of new and dense human and animal interactions is an important gap that planning and design scholars are positioned to fill. For these reasons, landscape scholars are well versed in the dynamics of the kinds of biotic, built, and social landscape patches that result in new and pathogenic viruses. However, since pandemics include ecological disturbances, whereby an EID is introduced to a new and non-proximal place, a socio-ecological landscape perspective is not helpful unless it envisions how one local landscape is connected to other local landscapes, a framework underlying the field of ecology. In other words, a policy relevant theoretical framework must be more refined than simply recommending universally closed borders, and account for precisely how a regional and localized epidemic becomes a global pandemic. It should provide a more comprehensive framework for charting the upstream and downstream sequences of events, and where they occur. Developing this conceptual framework, with an eye towards developing interventions to mitigate the risk of future pandemics is crucial because the stakes are so high in human and economic costs. COVID-19 has shown in stark terms the catastrophic impact that an EID can have on national and global supply chains, employment, and savings, to say nothing of the death toll and health impact. While these comprehensive impacts have yet to be tallied for the current pandemic, in 2017 the CDC estimated that global pandemics were likely to cost over $6 trillion over the next century, with an annualized expected loss of more than $60 billion for potential pandemics. However, investing $4.5 billion per year in building global capacities, it argued, could avert those catastrophic costs. Additionally, the Institute of Medicine (US) Forum on Microbial Threats (2009) and others suggested that some success had resulted from building platforms for improved virtual surveillance such as monitoring global news and health reports through artificial intelligence at the global scale. These policy approaches are, by definition, limited to once a pandemic has broken out and had time to affect large numbers of people, so they must be complemented by new tools for prediction and planning (e.g. Rubin et al., 2013, Kilpatrick and Randolph, 2016). Studies focused in this realm, however, have been scarce even though prediction and planning can help solve some of the difficult financial and political barriers to EID risk reduction by recommending evidence-based, and realistic prevention strategies (Spencer et al., 2021). While the phenomenon of EIDs may be new to planning scholars, it is important to note that zoonotic disease transmission has occurred throughout history, and microbiologists, medical doctors, and virologists have long sought to identify their origins based on the methodological tools of their disciplines. Increasingly, however, experts from these fields with conceptual and empirical tools centered on microbes, individual humans, and other a-spatial units of analysis have called for wider, landscape-level studies (e.g. Wilcox and Colwell, 2005, Neiderud, 2015). Here, I suggest that a planning and design research agenda for combining landscape analysis with global connectivity is needed to understand current and future pandemic outbreaks such as COVID-19, and thereby develop appropriate new policy tools. Murugesa (2020) has documented the growing interest of urban planners in the issue of EIDs, and Spencer et al. (2021) have provided a planning-oriented outline of how zoonoses – the phenomenon of animal viruses jumping from animal to humans – is at the center of EID origins. Preliminary empirical work has shown that large agglomerations of settlement and economic activity may be sources of these EIDs (e.g Spencer, 2013, Saksena et al., 2015), especially as they experience rapid urbanization and peri-urbanization. If this is indeed true, then with rapid urbanization of large parts of the developing world, especially the peri-urban regions of Asia and Africa, the risk conditions for the development of new pathogens such as Highly Pathogenic Avian Influenza, Ebola, and others, have been elevated (e.g. Institute of Medicine (US) Forum on Microbial Threats, 2009, Kontgis et al., 2014, Alirol et al., 2011, Alexander et al., 2015). Beyond these place-based investigations, a fewer number of scholars have emphasized global connectivity factors that connect distant communities as key drivers turning regional epidemics into global pandemics (e.g. Martin and Boland, 2018, Gendreau and DeJohn, 2002, Ryan et al., 2002). This dearth of studies of the origins of EIDs that usefully lead to new policy alternatives, therefore, requires an understanding of the upstream and downstream relationships from point A, where a zoonotic event of virus reassortment (Spencer et al., 2021) in a limited number of individuals living close to one another, to point Z, whereby hundreds of millions are infected globally. It is only a framework that integrates complex local landscape interactions and maps of planetary routes that connect these landscapes that can answer the simple question posed above. EIDs have been part of human populations for thousands of years, and there are surely many new zoonotic diseases that never become epidemics, let alone global pandemics. These new infectious disease outbreaks have generally been limited in geographic scope due to the new pathogens’ transmission through local populations until these host human groups were no longer viable hosts for the virus due to high mortality rates. Pandemics, however, disrupt these dynamics with relatively low short-term mortality rates, which allows low- and asymptomatic carriers to cross over wide distances undetected. Thus, newly sick individuals from these endemic landscapes where the pathogen has gradually arisen interact in new – non-proximal – landscapes where protection (both immunologic and social) has not had a chance to build up defenses. Pandemic EIDs are a function not just of dense local ecologies, but also the connectivity of these diverse ecologies across extensive distances that can introduce EID viruses into new landscape patches that have not been gradually exposed to these virus strains as they have developed over time. 3 The global urban ecosystem: revisiting the HIV/AIDS and pandemic influenza EIDs In prior work, I have extensively described a “global urban ecosystem” that characterizes the contemporary movement goods, people, ideas, and even viral micro-organisms as a global network of urban regions connected for both economic and cultural reasons (Spencer, 2014). This conceptual framework resting on the idea that global connectivity is comprised of interacting cities and regions (Spencer, 2014, pp. 7-38) is premised on ecological thinking that defines both “patches” and “corridors” (e.g. Henein and Merriam, 1990, Taylor et al., 1993), and associated principles of punctuated equilibrium (e.g. Grime, 1973, Connell, 1978). I argue here that EIDs are generated in specific types of landscape patches, and are transmitted globally through unregulated transportation corridors that “teleconnect” (Seto et al., 2012) geographically discontinuous landscape patches. It is through this GUE of landscape patches and teleconnecting transportation corridors that historic EIDs have turned regional epidemics into global pandemics. This framing of the GUE consolidates empirical work on the ecology and science of cities and regions (e.g. Bettencourt, 2013, Corning, 2002) and that of teleconnections that documents the direct cause and effect relationships of specific events in distant locales (e.g. Seto et al., 2012, Baird and Fox, 2015, Baird et al., 2019) into a shared conceptual framework. In sum, dense biotic and built landscapes gradually evolve and transform based on the human and animal individuals and communities that occupy them. This gradual evolution and transformation is periodically shaken by “disturbance” events that bring new, non-gradual human and animal individuals and communities into close proximity. As with periodic ecological disturbance events like wildfires and invasive species, viral disturbances can have devastating impact that result in deadly epidemics that transform local communities. Thus, while microscopic zoonotic transmission is necessary, it is not sufficient for creating a pandemic. Rather, pandemics are the result of these frequent and locally-rooted zoonotic transmission events, combined with networks that can disseminate them rapidly across the planet. A scientifically-informed narrative of the two most recent pandemics at the same scale as COVID-19 – Human Immunodeficiency Virus/Acquired Immunodeficiency Syndrome (HIV/AIDS) and pandemic influenza – can illustrate how the GUE operates. Before illustrating how it operates, however, it is important to recognize that even a preliminary and incomplete theory of how pandemics evolve can be essential for identifying ways to mitigate disaster. Taking a landscape-level approach to explaining such complex phenomena as global pandemics, even decades later, can provide only circumstantial evidence. Even the most commonly understood explanations cannot be definitively proven. Nevertheless, useful perspectives based on scientifically-informed narratives can change current and future policy even though evidence may remain largely circumstantial (Vance, 2019). Thus, a GUE level analysis of the HIV/AIDS and pandemic influenza crises may provide useful circumstantial evidence resulting in broad planning and policy guidelines that are simply good things to do, even though the “scientific answers might still be open.” Following Vance (2019), I argue here, there is enough “uncertainty in our [current] narratives” of EIDs to warrant a detailed and wide-ranging exploration of HIV/AIDS and pandemic influenza from the landscape perspective. 4 HIV/AIDS Given that the HIV/AIDS crisis became a pandemic over 40 years ago, it is remarkable how little planning research has engaged with the subject (Takahashi and Smutny, 2001, Takahashi, 1997), even though studies of HIV (Talman, Bolton, and Walson, 2013) have argued forcefully that urbanization and other landscape-level processes gave rise to it. From the public health perspective, Talman, Bolton, and Walson (2013) have posited what they call a “syndemic” framework to explain the evolution of HIV as a global pandemic. In sum, HIV/AIDS is the result of regularized human-simian interactions, regional EID spread through urbanization, transportation and unprotected sex; and global spread through travel, tourism and international aid. Today, 40 years after HIV/AIDS became known as a global pandemic, it is generally accepted that the disease has geographic origins in central Africa and biological origins in apes’ Simian Immunodeficiency Virus (SIV). After decades of ambiguity on the precise origins of the disease and numerous possible explanations of how it arose, today there is general consensus that the regional origins of the HIV pandemic center around Kinshasa, based on virologic studies establishing its presence in the region over 80 years ago (Fehervari, 2018). Specifically, David Ho discovered the first plasma sample to contain the HIV virus in the Central African region dating from 1959 (as cited in Fehervari, 2018), thereby definitively establishing evidence that HIV was present in humans in and around Kinshasa, Zaire as early as the 1950s. Beatrice Hahn studied wild chimpanzees in the region, finding that by around the year 2000 they served as hosts for an SIV virus almost identical to HIV. Related studies confirmed (Sharp and Hahn, 2011, Sharp et al., 2001) that the biological origin of the HIV virus was certainly among the SIV of the pan troglodyte subspecies of chimpanzee common in southern Cameroon, and that the jump from this group to humans occurred before 1940. Genetic analysis suggests that the first case of the HIV in Kinshasa may have been even earlier, between 1909 and 1930 (Gryseels et al., 2020, Faria et al., 2014). These studies of specific landscapes appear to tell a consistent story regarding the geographic origins of the primary strain of HIV that went global, a story that begins with the patterns of human/animal interactions surrounding Kinshasa, Zaire in the first half of the 20th Century. This historic virologic analysis among animals and humans in a single regional landscape in central Africa aligns with what has been found regarding the global distribution of HIV strain diversity. HIV is categorized into several subtypes, and each arose from independent cross-species infection events (Sharp & Hahn, 2011). In other words, HIV/AIDs does not originate in a single chance event; instead, numerous related but diverse viruses arose independent of one another in a shared geographic “landscape patch” (Hemelaar et al., 2019). Their proximity was the result of common regional environmental and human behavioral conditions, leading to a critical mass of events and exposures among humans. Because of this regularity, the studies suggest, it is clear that the HIV class of EIDs arose from complex regional dynamics rather than as the result of a single unique event. Hemelaar et al. (2019) reviewed empirical findings reported from 1990 to 2015, and mapped the drastic reduction in genetic variety of HIV strains outside of central Africa, arguing that it is smaller-n events that spread the disease rather than widespread community infection of the wide range of virus subtypes. Specifically, they argue that the HIV-C strain is almost exclusive in its dominance in South Africa, Ethiopia and India, and that HIV-B is almost as exclusively dominant in Western and Central Europe, North America, the Caribbean, and Central and South America. Thus, a landscape patch and transportation corridor approach strongly suggests a corridor connecting the local and regional landscape patch of central Africa in directions south and east, and a separate corridor connecting it west to Europe and the Americas led to two geographically distinct pandemics of the same virus. Again, while difficult to prove definitively, this compelling narrative outline can usefully shed light on the question of where HIV/AIDS came from by defining one pathway east and a different one west. The scientific evidence suggesting that HIV was endemic to the region surrounding Kinshasa is clear. To understand how this epidemic became a pandemic, however, it is then important to understand the level of connectivity between Kinshasa and other regional and global points. Epidemiologists suggest that diamond mine workers from Kenya, Zimbabwe, Uganda and Tanzania were significant vectors carrying the particular HIV strain back to their home countries by the 1970s (e.g. Faria et al., 2014, Dalai et al., 2009, Tully and Wood, 2010). From there, viral genetic evidence suggests, that the 1980s and 1990s saw the HIV-C strain (the most common version in South Africa and India) introduced into South Africa from these and other neighboring countries; moreover, given that South Africa was simultaneously a major destination for regional workers, as well as connected globally to economies outside of Africa, the conditions for turning an epidemic of central Africa into a global pandemic was enabled by a country with modern transportation infrastructure and global partnerships (Wilkinson et al., 2016). At the same time, HIV-B, which is the dominant strain in the Americas, is seen to have taken a different pathway that connects the central African region to global populations along a different transportation corridor of movement. Worobey Cox & Gill (2019) analyzed numerous cases of HIV samples and compared them with 117 from other parts of the world, estimating that it is 99% likely that Haiti served as the entry point for HIV in the United States. Moreover, the fact that Haiti shows the widest variety of HIV-B subtype of anywhere in the Americas, also suggests that it served as a node connecting central Africa to North and South America (Gilbert et al., 2007). Some outline the important role that the United Nations Organization in the Congo from 1960 − 1964 (ONUC) played in recruiting French-speaking professionals (doctors, engineers, and others) from Haiti to replace the departing Belgian colonial administrators (Jackson, 2014). Serving in this role for up to four years provided an adequate opportunity for Haitian professionals to intermingle in large numbers and over relatively long periods of time with native Congolese, thereby serving as unwitting points of entry for HIV to the Americas upon their return home to Haiti. Thus, these professionals facilitating decolonization in Africa may have been a key vector that allowed the HIV regional epidemic to connect the central African region to a non-proximal, global node. By 1966, HIV-B was established in Haiti and subsequently spread to the United States by around 1969, even though it did not become a commonly understood disease until much later. In sum, over 90 percent of HIV cases in South Africa, India, and Ethiopia are subtype C, whereas nowhere else in Africa does a single subtype comprise greater than 50 percent, and subtype B makes up well over 75% of the cases throughout the Americas and Australia/New Zealand (Gartner, Rochea, Churchill, Gorry, & Flynn, 2020). This evidence of distinct eastern and western teleconnecting transportation corridors originating from a shared patch in central Africa clearly suggest the importance of global corridors connecting regional and local socio-ecological landscapes. This story of landscape patches and teleconnecting transportation corridors is based on circumstantial evidence, pieced together through numerous empirical studies of the natural and health sciences, describes the key aspects of dynamic locality-based transformations in “mixing-bowl” like conditions at the animal/human interface, combined with growing regional and global network connections. Earlier, I posed the question “where did COVID-19 come from?” This is a question that remains imprecise even today for the HIV/AIDS pandemic. Yes, the virus itself did originate in the patch of central Africa, along with numerous other related viruses. But why did the type C virus dominate in India and type B in the Americas? These are the implied questions that cannot be answered by a single geographic identifier, but lie at the heart of preparing for a next pandemic. Part of the difficulty of developing policy alternatives for rare events like pandemics is the dearth of case studies. With the cases of recorded and well documented pandemics in the single digits, what kinds of lessons can be drawn from HIV/AIDs? So far, I have asserted that the GUE perspective of interconnected landscape patches and teleconnecting transportation corridors is a useful framework for shedding light on the origins of EIDs that can usefully lead to new policy alternatives and provided a narrative history of one known and studied pandemic to illustrate the case. This circumstantial evidence can be significantly strengthened by investigating the science and theories regarding the origins of pandemic influenza. Unlike the HIV/AIDs pandemic, the influenza pandemic of 1918 has been exhaustively summarized in book-length, science-based narratives that already point towards the importance of the GUE, but without naming it as such (e.g. Barry, 2005, Kolata, 1999). 5 Pandemic human influenza As an EID that has occupied the attention of health policy makers and researchers for over 40 years, especially in a time of advanced research methods and support, there is a rather large body of empirical research and evidence on the origins and dissemination of HIV/AIDS. However, the prior global pandemic of flu in the early part of the 20th Century occurred not only before exponential growth in the capacities of the biomedical and social sciences, but also during a period of tremendous global turmoil that surely limited virtually every nation’s capacity to focus on the disease. Nevertheless, a retrospective examination of the origins, dissemination, and likely causes of the 1918 “Spanish” Flu can pinpoint some very likely truths about the disease and pandemic that also illustrate the power of thinking in landscape terms that focus simultaneously on complex human-ecological patches connected globally by transportation networks enabling immediate interaction over non-proximal spaces. In his 2004 book The Great Influenza: The story of the deadliest pandemic in history, historian John Barry traces the evolution of the influenza virus through the global turmoil of World War I and its worldwide impact as it killed over 50 million individuals in the course of two years, 675,000 in the United States alone (CDC, 2020). As a scientifically informed, popularly oriented narrative, Barry’s work weaves stories of key individuals into the disjointed facts that are known about the virology of the influenza virus and its impacts. For my purposes of illustrating the importance of regional design and planning through a lens of dense landscape patches of human and animal interactions and teleconnecting transportation corridors placing non-proximal landscapes into immediate and direct contact, his description of the origins of the 1918 Influenza pandemic disease is interesting. His summary of what is known about the origins and spread of the disease fits neatly with the argument presented here:Epidemiological evidence suggests that a new influenza virus originated in Haskell County, Kansas, early in 1918. Evidence further suggests that this virus traveled east across the state to a huge army base, and from there to Europe. Later, it began its sweep through North America, through Europe, through South America, through Asia and Africa, through isolated islands in the Pacific, through all the wide world (p. 92). Using the metaphors of “swarms” and “tinderboxes”, Barry goes on to describe the mixing bowl characteristics of central Kansas at the turn of the 19th Century. At this time, the United States was undergoing a rapid industrialization of its economy, and the exponential growth of urban centers associated with it. In 1870, he points out, the United States’ population of 40 million people lived predominantly – almost 75% – in small towns or on farms; by 1914, the population had reached 105 million and several years later more than 50% of them lived in urban areas (p. 122). Naturally, the growth of cities at the time required increased agricultural productivity, and an explosion of industrial agriculture helped fuel urban growth during this period. Barry describes Haskell County as a place where “…land, crops and livestock were everything, and the smell of manure meant civilization. Farmers lived in close proximity to hogs and fowl, with cattle, pigs and poultry everywhere.” (p. 91), This history aligns with contemporary theories on today’s deadly avian influenzas are the result of these “mixing bowl” landscapes characterized by the dense interactions of humans and animals – and their viruses – without much attention from planning and design. Reading Barry’s scientifically informed narrative of the geography of Haskell County and the argument that it is the origin of the 1918 pandemic, it is difficult not to imagine the contemporary urbanization and agricultural intensification dynamics of regions in the 21st Century such as China, Viet Nam, Thailand, Indonesia and elsewhere (as described in Spencer et al., 2020, Wilcox and Colwell, 2005, Kapan et al., 2006). As the evidence of HIV/AIDs origins suggested, however, the creation of new viruses through zoonotic transmission, while a health threat locally, does not constitute a pandemic on its own. Rather, the connections across wide distances that position infected humans in close proximity with others who have a very different profile of immunity is what Barry calls the “tinderbox.” In 1918 these local virus dynamics combined with the exponential growth of global military enlistment and travel associated with World War I to disseminate a locally generated zoonotic disease across the globe. Farmers and their sons from Haskell County enlisted at nearby Camp Funston near Manhattan KS, joining thousands of other young men in cramped, often unsanitary living conditions – conditions replicated at US and other military camps around the world – thereby creating the perfect conditions for the 1918 Influenza to spread globally. This macro-level understanding of the origins and spread of the 1918 Influenza is not simply historical speculation informed by a limited body of biological and public health evidence. Rather, evolutionary biologists have also begun to reconsider the 1918 Influenza in the context of rapidly changing landscapes of urbanization and farming. Worobey, Cox and Gill (2019) have critically evaluated Barry’s claim as to the origins in Haskell County, in the context of two other origin possibilities in France and China, finding some contradictory evidence of simultaneous infection in New York City. They conclude, however, that Barry’s story has a stronger evidentiary basis than the other two origin stories, whether or not Haskell County can be definitively shown to be the origins. It appears fruitful, therefore, to developing a planning-relevant understanding of EIDs from this urbanization perspective. On this point, Worobey’s research is clearer. According to Worobey, “[w]hat we’re seeing is maybe our domesticated animals like ducks and chickens, they might be this huge pool of vulnerable hosts that viruses can get into and amplify and then you can have spillover from domestic into the wild” (as recounted by Barchfield, 2014). His research finds that there were major outbreaks of horse influenza and avian influenza in Haskell County during the years immediately preceding the 1918 flu, suggesting that influenza had long found receptive hosts in large-scale domesticated animals, and that the 1918 Influenza did not simply appear from nowhere as the result of a one-off event. Instead, this history of prior circulation among animal stock – as opposed to wild species, as was the case for HIV/AIDs – suggests to Worobey and his colleagues that the human-livestock interface is the most likely origins of the 1918 human influenza. Moreover, he argues, it suggests a more logical origin story than antigenic drift within humans, which is sometimes used to explain the origins of pandemic influenza and other EIDs. In other words, the implications of Worobey’s genetic analysis broadly concur with Barry’s “most-likely” explanation that it was the intensification of farming practices and the dense proximity of humans and livestock that likely led to the 1918 pandemic, whether the specific case happened in Haskell County or not. Once developed in these dense patches of rapid socio-economic, cultural, and ecological transition, it is not hard to see how the influenza virus quickly transmitted across the military camps and communities that operated at a global scale during this advanced stage of World War I. Thus, just as colonialism/decolonization and growing global trade links to Central Africa served as an unwitting corridor for pandemic, World War I also helped create a pandemic out of a regional epidemic. Since 1918, the human influenza virus has been an annual infection spreading widely among human populations without anywhere near the deadly impacts of 1918. Mostly, it has been a nuisance dealt with through vaccination, masks, and isolation. Periodically, however, a new virulent strain of influenza originates in a livestock and jumps to humans with deadly consequences. In the 2000s, for example, Highly Pathogenic Avian Influenza (HPAI) threatened to lead to major epidemics and possibly a pandemic due in large part to urbanization and agricultural practices (Saksena et al., 2015, Finucane et al., 2014). During the 2004–05 outbreak of avian influenza in Viet Nam, for example, 119 people were infected with the H5N1 virus, and 59 died—an alarming mortality rate of nearly 50 percent. Nearly all of these infections were the result of direct contact with diseased chickens or ducks, and the virus did not have time to evolve into a strain able to transmit easily between humans. Had the virus spread more widely within the human population, public health officials estimate that thousands more would have died. Fortunately, this epidemic EID strain failed to evolve into a pandemic. Empirical analyses of this more recent strain of influenza do show, however, that the zoonotic influenzas occurred in the kinds of rapidly urbanizing and agriculturally intensifying landscapes that Barry describes in Haskell County (Saksena et al., 2015, Spencer et al., 2020, Finucane et al., 2014, Spencer, 2013). Clearly the landscape conditions that facilitated the evolution of the original HIV/AIDs and influenza viruses remain throughout large portions of the world. In today’s connected world, these regional dynamics are certainly connected to global networks that might transform regional epidemics into pandemics even after COVID-19 has subsided. 6 Planning for global pandemics requires outlining the global landscape: a research and practice strategy for EIDs This condensed history of the origins and dissemination of HIV/AIDS and pandemic influenza outlined above, while admittedly circumstantial based on limited empirical evidence, does clearly point to critical sequences of events that happened in transitional, developing areas at the interface of animals (both wild and domesticated) and people. Moreover, the regional and global links that allowed individuals – some of whom were symptomatic and asymptomatic carriers of EIDs – across both moderately and extremely distant spaces without any systematic surveillance or regulation turned epidemics into pandemics. What can we learn from HIV/AIDs and pandemic influenza by asserting the narrative of the GUE? The circumstantial evidence described above should give planners and designers the confidence to believe they can chart the most likely patches of zoonotic EIDs and their corridors of likely spread. An empirically informed GUE map, thus, would provide guidance for the most effective places to look for and mitigate these developing regional EIDs before they become pandemics. 7 What is the landscape where pandemic planning should happen? the global urban ecosystem It is in this realm of developing risk mitigation strategies for future pandemics that the metaphor of a GUE can be useful, even if the science cannot be definitively proven. Because there will always be uncertainty in a narrative explanation of a problem on such a global scale, looking to other managed systems of patches and corridors can be useful for policy and planning design. The Global Urban Ecosystem described earlier is best understood as a kind of global version of an urban subway system (Spencer, 2014, pp. 29–30), whereby non-proximal “neighborhoods” are connected through a standardized system of linear pathways that make travel to unknown areas predictable and navigable. However, unlike a formal transportation infrastructure such as a metro system or a flight network, the Global Urban Ecosystem is governed by a diffuse and disjointed set of global regulations on the mobility of people, goods and services, combined with numerous cultural and linguistic norms that allow it to function. While more complex than a unilaterally constructed and managed system, understanding the dynamics and learning patterns inherent to this type of global system of connected landscapes – each of which serves as a complex mixing bowl of human and animal viruses – is essential for tackling the human-driven challenges of EIDs. Fortunately, the metaphor of a metropolitan transportation system provides more than just a conceptual framework for scholars; as a human developed and managed system, it provides a set of policy and planning alternatives for practicing planners and designers to consider, evaluate, and adapt. Above, I have described a global system of regional and local landscape patches connected by global transportation corridors, and the importance of understanding this system as it relates to EIDs. This defined system, however, is not governed by any planning institution and thus requires further specification related to what kinds of actions might be possible to limit exposure to new EIDs. A metropolitan transportation system is a small version of how intergovernmental entities create and manage a system of neighborhood patches with regional teleconnecting corridors. This managed system not only achieves transportation goals, but also mitigates a range of risks and contextual factors that facilitate its safe and effective operations. Managing transportation systems requires surveillance, stations, a trained security and workforce, and a coordinated schedule and emergency management plan. If a disturbance breaks out at one station, for example, numerous mathematical algorithms chart new connection routes to minimize passengers transiting through that station, while suspending routes that have that station as their destination. Similarly, most modern transit systems have some form of ticket tracking for knowing the prices, locations, and durations of millions of individual trips each day, consistently analyzing them for the purposes of deploying resources where they are most needed to optimize the system for cost efficiency, time efficiency, or some other goal. Moreover, a core function of the system is to define and incentivize appropriate behavior through signs and checkpoints not only in the system itself, but in the neighborhoods patches that it connects. The former is self-evident, but the latter includes neighborhood directional signs, outlets for ticket purchasing, zoning for appropriate proximate land uses, and connecting infrastructure to local transportation networks (e.g. “last-mile” connections) and residential practices (e.g. Transit-Oriented Development). Had the technology been available at the time, and a consensus to optimize health risk, would planners and designers have been able to track elevated numbers of low-level fevers departing from Haskell County, or high frequencies of arrivals of “underweight” travelers to Haiti. If so, would they have gotten a sense that some kind of health risk was elevated along one of the regularly traveled corridors? Conversely, starting from the standpoint of a landscape patch, if an EID is discovered, its risk of becoming a pandemic is largely defined by the transit node’s catchment area, similar to the conventional ¼ mile walking radius of a transit station. The existence of this kind of “landscape patch”/“teleconnecting transportation corridor” system prompts a number of interesting speculative questions, some of which may be valuable for future policies regarding EIDs and global pandemics. Such new kinds of surveillance other kinds of planning prompts major ethical questions, and the goal of such surveillance, as with a transit system, is not to penalize any given individual in one of the patches or corridors, but rather a tool for deploying limited resources in ways that optimize the goal of minimizing health risks. Unlike during the HIV/AIDs and pandemic influenza periods, such kinds of technologies do currently exist or are easily envisioned for the near future if demand is sufficient. As with the rapidly growing literature on “smart cities,” (e.g. Bettencourt, 2013), the thorny ethical questions of surveillance and individual rights can be more difficult to navigate than the technology itself. This institutional question can also be partly answered by an examination of a transit system. Beyond the day-to-day management, many metropolitan transit systems have a public governing board and a cross-jurisdictional financing mechanism that allow them to coordinate across municipal and state lines, as well as distribute the costs to those using services appropriately. These simple illustrations show that thinking of the GUE’s role in pandemics may be useful for developing future plans and designs to mitigate the risk of EIDs. Health experts have, of course, made recommendations to manage the growing risk of EIDs, such as creating surveillance systems to rapidly detect and report cases, developing an international laboratory network to accurately identify local causes of illness, training a workforce to identify, track, and contain outbreaks at their sources, and coordinating emergency management systems to mount effective responses. Placing these isolated recommendations into a more systematic institutional framework defined by the landscape patches and teleconnecting corridors inherent to the GUE, I argue, would allow these isolated efforts to be targeted more efficiently and sustainably. Clearly the GUE is much more complex than a metropolitan transit system. Nevertheless, our planning and design experience with such connected systems of landscape patches can provide much needed organizational framework to usefully define policy responses to the growth and consequences of EIDs. ==== Refs References Alexander K.A. Sanderson C.E. Marathe M. Lewis B.L. Rivers C.M. Shaman J. …Akogun O.B. What factors might have led to the emergence of ebola in west Africa? PLoS Neglected Tropical Diseases 9 6 2015 e0003652 10.1371/journal.pntd.0003652 26042592 Alirol E. Getaz L. Stoll B. Chappuis F. Loutan L. Urbanisation and infectious diseases in a globalised world The Lancet – Infectious Diseases 11 2 2011 131 141 10.1016/S1473-3099(10)70223-1 21272793 Baird I.G. Fox J. How land concessions affect places elsewhere: Telecoupling, political ecology, and large-scale plantations in southern Laos and northeastern Cambodia Land 4 2 2015 436 453 10.3390/land4020436 Baird I.G. Noseworthy W. Tuyen N.P. Ha L.T. Fox J. Land grabs and labour: Vietnamese workers on rubber plantations in southern Laos Singapore Journal of Tropical Geography 40 1 2019 50 70 10.1111/sjtg.2019.40.issue-110.1111/sjtg.12261 Barchfield, V. (2014, March 21). UA professors trace[s] origins of flu. Arizona Public Media Retrieved from https://news.azpm.org/s/17788-ua-professors-traces-origins-of-flu/. Barry J.M. The great influenza: The epic story of the deadliest plague in history 2005 Penguin Books New York Bettencourt L.M.A. The origins of scaling in cities Science 340 6139 2013 1438 1441 10.1126/science.1235823 23788793 Boyce M.R. Katz R. Standley C.J. Risk factors for infectious diseases in urban environments: A systematic review and critical appraisal of evidence Tropical Medicine and Infectious Disease 4 4)(123 2019 1 18 10.3390/tropicalmed4040123 Centers for Disease Control and Prevention (CDC). (2017). One health zoonotic disease prioritization for multi-sectoral engagement in Uganda. Paper presented at the Kampala, Uganda. 1-19. Retrieved from https://www.cdc.gov/onehealth/what-we-do/zoonotic-disease-prioritization/completed-workshops.html. Centers for Disease Control and Prevention (CDC). (2020a). Reports from complete workshops on one health zoonotic disease prioritization. Retrieved from https://www.cdc.gov/onehealth/basics/index.html. Centers for Disease Control and Prevention (CDC). (2020b). Why it matters: The pandemic threat. Retrieved from https://www.cdc.gov/globalhealth/healthprotection/fieldupdates/winter-2017/why-it-matters.html. Connell J.H. Diversity in tropical rain forests and coral reefs Science 199 4335 1978 1302 1310 10.1126/science.199.4335.1302 17840770 Corning P.A. The re-emergence of 'emergence': A venerable concept in search of a theory Complexity 7 6 2002 18 30 10.1002/cplx.v7:610.1002/cplx.10043 Cudahy B.J. Box Boats: How container ships changed the world 2006 Fordham University Press New York Dalai, S. C., de Oliveira, T., Harkins, G. W., Kassaye, S. G., Lint, J., Manasa, J., Katzenstein, D. (2009). Evolution and molecular epidemiology of subtype C HIV-1 in Zimbabwe. Aids, 23(18), 2523-2532. doi:10.1097/QAD.0b013e3283320ef3. Daszak, P., Eisen, J., Gaunt, E., Morse, S., & Woolhouse, M. (2009). Infectious disease emergence: Past, present, and future. In D. A. Relman, M. A. Hamburg, E. R. Choffnes, A. Mack, M. J. Blaser, S. J. Brickner, M. Worth (Eds.), Microbial evolution and co-adaptation: A tribute to the life and scientific legacies of Joshua Lederberg (pp. 193-272). Washington, D.C.: The National Academies Press. Retrieved from https://www.ncbi.nlm.nih.gov/books/NBK45714/#top. Faria N.R. Rambaut A. Suchard M.A. Baele G. Bedford T. Ward M.J. …Lemey P. The early spread and epidemic ignition of HIV-1 in human populations Science 346 6205 2014 56 61 10.1126/science:1256739 25278604 Fehervari Z. November 28). Origin story Retrieved from Nature Milestones 9 2018 https://www.nature.com/articles/d42859-018-00008-6 Finucane M.L. Nghiem T. Saksena S. Nguyen L. Fox J. Spencer J.H. Thau T.D. An exploration of how perceptions of the risk of avian influenza in poultry relate to urbanization in Vietnam Ecohealth 11 2014 73 82 10.1007/s10393-014-0906-2 24488189 Gartner M.J. Rochea M. Churchill M.J. Gorry P.R. Flynn J.K. Understanding the mechanisms driving the spread of subtype C HIV-1 EBioMedicine 53 2020 102682 Gendreau M.A. DeJohn C. Responding to medical events during commercial airline flights New England Journal of Medicine 346 14 2002 1067 1073 10.1056/NEJMra012774 11932475 Gilbert M.T.P. Rambaut A. Wlasiuk G. Spira T.J. Pitchenik A.E. Worobey M. The emergence of HIV/AIDS in the Americas and beyond Proceedings of the National Academy of Sciences of the United States of America 104 47 2007 18566 18570 10.1073/pnas.0705329104 17978186 Grime J.P. Competitive exclusion in herbaceous vegetation Nature 242 1973 344 347 Retrieved from https://www.nature.com/articles/242344a0.pdf?origin=ppub Gryseels S. Watts T.D. Kabongo Mpolesha J.-M. Larsen B.B. Lemey P. Muyembe-Tamfum J.-J. …Worobey M. A near full-length HIV-1 genome from 1966 recovered from formalin-fixed paraffin-embedded tissue Proceedings of the National Academy of Sciences of the United States of America 117 22 2020 12222 12229 10.1073/pnas.1913682117 32430331 Hemelaar J. Elangovan R. Yun J. Dickson-Tetteh L. Fleminger I. Kirtley S. …Zhang R. Global and regional molecular epidemiology of HIV-1, 1990–2015: A systematic review, global survey, and trend analysis Infectious Diseases 19 2 2019 143 155 10.1016/S1473-3099(18)30647-9 Henein K. Merriam G. The elements of connectivity where corridor quality is variable Landscape Ecology 4 2-3 1990 157 170 10.1007/BF00132858 Institute of Medicine (US) Forum on Microbial Threats. (2009). Infectious disease emergence: Past, present, and future. Paper presented at the Microbial Evolution and Co-Adaptation: A Tribute to the Life and Scientific Legacies of Joshua Lederberg: Workshop Summary, Washington, D.C. doi:10.17226/12586. Jackson R.O. The failure of categories: Hatians in the United Nations organization in The Congo, 1960–64 Journal of Haitian Studies 20 1 2014 34 64 Kapan D.D. Bennett S.N. Ellis B.N. Fox J. Lewis N.D. Spencer J.H. Wilcox B.A. Avian influenza (H5N1) and the evolutionary and social ecology of infectious disease emergence EcoHealth 3 3 2006 187 194 10.1007/s10393-007-0095-3 Kilpatrick, A. M., & Randolph, S. E. (2016). Drivers, dynamics, and control of emerging vector-borne zoonotic diseases. In A. Mack, J. Fletcher, R. Khabbaz, D. Pfeiffer, D. Rizzo & M. Worth (Eds.), Global health impacts of vector-borne diseases (pp. 182-201). Washington, D.C.: National Academies Press. doi:10.17226/21792 Retrieved from https://www.ncbi.nlm.nih.gov/books/NBK355538/ (Original work published in 2012 in The Lancet; reprinted with permission). Kolata G. Flu: The story of the great influenza pandemic of 1918 and the search for the virus that caused it 1999 Farrar, Straus, and Giroux New York Kontgis C. Schneider A. Fox J. Saksena S. Spencer J.H. Castrence M. Monitoring peri-urbanization in the greater Ho Chi Minh City metropolitan area Applied Geography 53 2014 377 388 10.1016/j.apgeog.2014.06.029 Martin G. Boland M. Planning and preparing for public health threats at airports Globalization and Health 14 28 2018 1 5 10.1186/s12992-018-0323-3 29310698 Murugesa, V. (2020). How can Urban Planners Address Emerging Zoonoses? A Scoping Review of Recommended InterventionsUrbana, XXI. Retrieved from http://www.urbanauapp.org/ 10.47785/urbana.4.2020. Neiderud C. How urbanization affects the epidemiology of emerging infectious diseases Infection Ecology and Epidemiology 5 1 2015 1 9 10.3402/iee.v5.27060 Rubin C. Myers T. Stokes W. Dunham B. Harris S. Lautner B. Annelli J. Review of institute of medicine and national research council recommendations for one health initiative Emerging Infectious Diseases 19 12 2013 1913 1917 10.3201/eid1912.121659 24274461 Ryan E.T. Wilson M.E. Kain K.C. Illness after international travel New England Journal of Medicine 347 7 2002 505 516 10.1056/NEJMra020118 12181406 Saksena S. Fox J. Epprecht M. Tran C.C. Nong D.H. Spencer J.H. …Eugenin E.A. Evidence for the convergence model: The emergence of highly pathogenic avian influenza (H5N1) in Viet Nam PLoS ONE 10 9 2015 e0138138 10.1371/journal.pone.0138138 26398118 Seto K.C. Reenberg A. Boone C.G. Fragkias M. Haase D. Langanke T. …Simon D. Urban land teleconnections and sustainability Proceedings of the National Academy of Sciences of the United States of America 109 20 2012 7687 7692 10.1073/pnas.1117622109 22550174 Sharp P.M. Bailes E. Chaudhuri R.R. Rodenburg C.M. Santiago M.O. Hahn B.H. The origins of acquired immune deficiency syndrome viruses: Where and when? Philosophical Transactions of the Royal Society B, Biological Sciences 356 1410 2001 867 876 10.1098/rstb.2001.0863 Sharp P.M. Hahn B.H. Origins of HIV and the AIDS pandemic Cold Spring Harbor Perspectives in Medicine 1 1 2011 1 22 10.1101/cshperspect.a006841 Spencer J.H. The urban health transition hypothesis: Empirical evidence of an avian influenza Kuznets curve in Vietnam? Journal of Urban Health 90 2 2013 343 357 10.1007/s11524-012-9719-1 22798150 Spencer J.H. Globalization and urbanization: The global urban ecosystem 2014 Rowman & Littlefield Publishers Lanham, MD Spencer J.H. Finucane M. Fox J. Saksena S. Sultana N. Emerging infectious disease, the built landscape, and urban planning: Evidence on avian influenza in Viet Nam Landscape and Urban Planning 193 2020 1 14 10.1016/j.landurbplan.2019.103681 Spencer J.N.H. Marasco D. Eichinger M. Planning for Emerging Infectious Disease Pandemics: Definitions, the role of planners, and learning from the avian influenza outbreak of 2004–2005 Journal of the American Planning Association 2021 10.1080/01944363.2021.1930107 Takahashi L.M. The socio-spatial stigmatization of homelessness and HIV/AIDS: Toward an explanation of the NIMBY syndrome Social Science and Medicine 45 6 1997 903 914 10.1016/S0277-9536(96)00432-7 9255923 Takahashi L.M. Smutny G. Collaboration among small, community-based organizations: Strategies and challenges in turbulent environments Journal of Planning Education and Research 21 2 2001 141 153 10.1177/0739456X0102100203 Talman A. Bolton S. Walson J.L. Interactions between HIV/AIDS and the environment: Toward a syndemic framework American Journal of Public Health 103 2 2013 253 261 10.2105/AJPH.2012.300924 23237167 Taylor P.D. Fahrig L. Henein K. Merriam G. Connectivity is a vital element of landscape structure Oikos 68 3 1993 571 573 10.2307/3544927 Tully D.C. Wood C. Chronology and evolution of the HIV-1 subtype C epidemic in Ethiopia Aids 24 10 2010 1577 1582 10.1097/QAD.0b013e32833999e1 20539092 Vance M.A. Conflicting views in narratives on HIV transmission via medical care Journal of the International Association of Providers of AIDS Care 18 2019 1 9 10.1177/2325958218821961 Wilcox B.A. Colwell R.R. Emerging and reemerging infectious diseases: Biocomplexity as an interdisciplinary paradigm EcoHealth 2 4 2005 244 257 10.1007/s10393-005-8961-3 Wilkinson E. Rasmussen D. Ratmann O. Stadler T. Engelbrecht S. de Oliveira T. Origin, imports and exports of HIV-1 subtype C in South Africa: A historical perspective Infection, Genetics and Evolution 46 2016 200 208 10.1016/j.meegid.2016.07.008 Worobey M. Cox J. Gill D. The origins of the great pandemic Evolution, Medicine, and Public Health 2019 1 2019 18 25 10.1093/emph/eoz001 30805187 Wu T. Perrings C. Kinzig A. Collins J.P. Minteer B.A. Daszak P. Economic growth, urbanization, globalization, and the risks of emerging infectious diseases in China: A review Ambio 46 1 2017 18 29 10.1007/s13280-016-0809-2 27492678

0

PMC9754155

NO-CC CODE

2022-12-16 23:26:28

no

Landsc Urban Plan. 2021 Dec 15; 216:104242

utf-8

Landsc Urban Plan

10.1016/j.landurbplan.2021.104242

oa_other

==== Front Diabetes Res Clin Pract Diabetes Res Clin Pract Diabetes Research and Clinical Practice 0168-8227 1872-8227 Elsevier B.V. S0168-8227(21)00035-8 10.1016/j.diabres.2021.108682 108682 Article The impact of a prolonged lockdown and use of telemedicine on glycemic control in people with type 1 diabetes during the COVID-19 outbreak in Saudi Arabia Alharthi Sahar K. a1 Alyusuf Ebtihal Y. b1 Alguwaihes Abdullah M. b Alfadda Assim bc Al-Sofiani Mohammed E. bde⁎ a Department of Internal Medicine, College of Medicine, King Saud University, Riyadh, Saudi Arabia b Division of Endocrinology, Department of Internal Medicine, College of Medicine, King Saud University, Riyadh, Saudi Arabia c Obesity Research Center, College of Medicine, King Saud University, Riyadh, Saudi Arabia d Division of Endocrinology, Diabetes & Metabolism, The Johns Hopkins University, Baltimore, MD, United States e Strategic Center for Diabetes Research, College of Medicine, King Saud University, Riyadh, Saudi Arabia ⁎ Corresponding author at: Division of Endocrinology, Department of Internal Medicine, College of Medicine, King Saud University, Riyadh, Saudi Arabia. 1 Contributed equally to this work. 2 2 2021 3 2021 2 2 2021 173 108682108682 25 11 2020 3 1 2021 19 1 2021 © 2021 Elsevier B.V. All rights reserved. 2021 Elsevier B.V. Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. Background To minimize the spread of Coronavirus Disease-2019, Saudi Arabia imposed a nationwide lockdown for over 6 weeks. We examined the impact of lockdown on glycemic control in individuals with type 1 diabetes (T1D) using continuous glucose monitoring (CGM); and assessed whether changes in glycemic control differ between those who attended a telemedicine visit during lockdown versus those who did not. Materials and Methods Flash CGM data from 101 individuals with T1D were retrospectively evaluated. Participants were categorized into two groups: Attended a telemedicine visit during lockdown (n = 61) or did not attend (n = 40). Changes in CGM metrics from the last 2 weeks pre-lockdown period (Feb 25 - March 9, 2020) to the last 2 weeks of complete lockdown period (April 7–20, 2020) were examined in the two groups. Results Those who attended a telemedicine visit during the lockdown period had a significant improvement in the following CGM metrics by the end of lockdown: Average glucose (from 180 to 159 mg/dl, p < 0.01), glycemic management indicator (from 7.7 to 7.2%, p = 0.03), time in range (from 46 to 55%, p < 0.01), and time above range (from 48 to 35%, p < 0.01) without significant changes in time below range, number of daily scans or hypoglycemic events, and other indices. In contrast, there were no significant changes in any of the CGM metrics during lockdown in those who did not attend telemedicine. Conclusions A six-week lockdown did not worsen, nor improve, glycemic control in individuals with T1D who did not attend a telemedicine visit. Whereas those who attended a telemedicine visit had a significant improvement in glycemic metrics; supporting the clinical effectiveness of telemedicine in diabetes care. Keywords Telemedicine Type 1 Diabetes Lockdown Continuous Glucose Monitoring COVID-19 ==== Body pmc1 Introduction Maintaining a good glycemic control in people with type 1 diabetes (T1D) is often challenging during “ordinary” times; and becomes more challenging and important during times of uncertainty and physical and mental distress. Glucose levels in people with T1D are affected by many factors including levels of physical activity, eating patterns, physical and mental health, frequency of glucose monitoring, adherence to insulin therapy, and availability of an uninterrupted access to healthcare providers (HCPs) [1], [2]. The unprecedented Coronavirus disease 2019 (COVID-19) pandemic has caused abrupt and unplanned-for changes in almost all of these factors; and people living with T1D were among those impacted the most by this pandemic. Furthermore, many countries have implemented precautionary measures including closures of schools and non-essential businesses, cancellations of routine (non-urgent) clinic appointments, and lockdowns, to mitigate the spread of COVID-19 [3]. Saudi Arabia has implemented one of the longest and most strictly enforced nationwide lockdowns that lasted for a total of 3 months. Some parts of the countries had a complete lockdown (i.e. 24 hr/day) that lasted for over a month; whereas others had approximately 3 weeks of complete lockdown. The complete lockdown was preceded and followed by a total of 8 weeks of partial lockdown (i.e. at least 15 hrs/day) [4], [5]. Recent studies have shown a negative impact of lockdown on health behaviors in people with diabetes [6], [7], [8]. Therefore, it is hypothesized that a prolonged lockdown, such as the one implemented in Saudi Arabia, would likely result in worsening of glycemic control in people living with diabetes. In addition, many people with T1D lost access to their HCPs during the COVID-19 outbreak as a result of the lockdown and cancellation of routine clinic appointments. Though some patients with T1D were able to utilize telemedicine to restore communications with their HCPs during this difficult time, telemedicine was not available for the majority of people with T1D, particularly those who live in parts of the world where telemedicine is not well-established [9], [10]. Fortunately, the recent improvements in diabetes technologies, including the availability of continuous glucose monitoring (CGM) devices and blood glucose meters with remote data sharing capabilities, have made the transition to telemedicine during the COVID-19 outbreak a relatively smooth process for many clinics. We have recently described our experience with implementing a simplified Diabetes Telemedicine Clinic for people with diabetes living in Saudi Arabia, where diabetes is highly prevalent and telemedicine is not well-established. We showed that a Diabetes Telemedicine Clinic, utilizing simple technological tools widely available to most people with diabetes and HCPs such as smart phones and virtual communication applications, is not only feasible but also associated with high patients’ and physicians’ satisfaction [11]. The clinical effectiveness of such a simplified diabetes telemedicine clinic remains unknown, particularly when implemented in areas where telemedicine was not well-established prior to the COVID-19 pandemic. In this study, we evaluate the impact of a six-week lockdown on CGM glycemic indices in people with T1D. We also examine the clinical effectiveness of a simplified diabetes telemedicine clinic, that was rapidly implemented at the beginning of the COVID-19 outbreak in Saudi Arabia utilizing technological tools that are widely available for most people with T1D and HCPs. 2 Subjects and methods 2.1 Study design and participants A retrospective analysis of FreeStyle Libre CGM data from 101 individuals with T1D who attended the Specialized Diabetes Clinic at King Saud University Medical City in Riyadh, Saudi Arabia prior to the COVID-19 pandemic and had their data remotely shared with our clinic using Libreview, a web-based cloud system. Participants who did not have their CGM data shared with our clinic during these periods were excluded. The timeline of the COVID-19 outbreak and important dates related to the outbreak and lockdown in Saudi Arabia are shown in Fig. 1 . A nationwide closure of schools and non-essential businesses was ordered on March 10, 2020; followed by a nationwide partial lockdown (i.e. from 3 PM to 6 AM) starting on March 24, 2020 that was then upgraded to a nationwide complete lockdown (i.e. a 24-hr lockdown) from April 7 to April 23, 2020. After that, the lockdown was downgraded and became a partial lockdown again which lasted for almost two months (from April 24 to June 21, 2020). In this study, we examined changes in CGM metrics from the “Pre-lockdown” period (i.e. February 25 to March 9) to the “Complete Lockdown” period (i.e. April 7 to 20). The study was approved by the Institutional Review Board at King Saud University.Fig. 1 A Nationwide closure of schools and non-essential businesses was ordered on March 10, 2020; followed by a partial lockdown starting on March 24, 2020 that was upgraded to a complete lockdown from April 7 to April 23, 2020. Diabetes Telemedicine Clinic was implemented on March 10, 2020.. 2.2 Assessment of glycemic outcomes and covariates The following CGM metrics were evaluated in all the study participants both pre-lockdown and at the end of the complete lockdown period: Average glucose, coefficient of variation (CV), glucose management indicator (GMI), TIR (time in range 70–180 mg/dl), TAR (time above range > 180 mg/dl), TBR (time below range < 70 mg/dl), CGM active time, and number of daily scans. In addition, the following demographic and clinical data were collected from the electronic medical records: Age, sex, diabetes duration, body mass index, hemoglobin A1C prior to the lockdown, and mode of insulin therapy. 2.3 Telemedicine clinic visit We also collected data on whether the study participants attended a virtual visit with their diabetes care providers during the lockdown period or not; as this could be a potential predictor of changes in CGM metrics during lockdown. Study participants who had at least one virtual visit to our Diabetes Telemedicine Clinic during the lockdown period were considered in the “Telemedicine Visit” group; whereas, those who had no virtual visit during the lockdown were considered in the “No Telemedicine Visit” group. The detailed protocol of our Diabetes Telemedicine Clinic has been recently described in another paper [11]. Briefly, people with T1D were eligible to be seen in our Diabetes Telemedicine Clinic during the lockdown period if they had already been scheduled for a routine follow-up visit from prior to the pandemic and had any of the following: A1C of ≥ 9%, >1 reported hypoglycemic event/week, were planning to fast during the month of Ramadan, or requested to be seen virtually. All other people with T1D, who did not meet any of these criteria, were offered the option to postpone their clinic visit to 3 months. 2.4 Statistical analysis Variable distributions were examined using Shapiro-Wilk test and visual examination of histograms. The non-normally distributed continuous variables were presented as medians and interquartile ranges (IQR); whereas, categorical variables were presented as percentages. Baseline characteristics were compared between those who had a telemedicine visit versus those who didn’t have any visit during the lockdown period using Kruskal-Wallis test for the non-normally distributed continuous variables and chi-squared tests of homogeneity for categorical variables. The changes in CGM glycemic metrics within each study group over the six-week lockdown period were examined using Wilcoxon signed rank test. All significance testing was 2-tailed with α of 0.05, and data were analyzed using Stata Statistical Software (release 15). 3 Results 3.1 Baseline characteristics A total of 101 individuals with T1D were included in the study, of whom 55% were women and 60% attended a telemedicine visit during the lockdown period. The median age of the study participants was 23 years with a median diabetes duration of 7 years and hemoglobin A1C of 8.3% at baseline. No significant differences were noted in age, gender, diabetes duration, BMI, baseline hemoglobin A1C, total daily dose of insulin, mode of insulin therapy, or prevalence of diabetic complications between those who attended a telemedicine visit during the lockdown period and those who did not (all p values > 0.05) (Table 1 ). However, those who attended a telemedicine visit during the lockdown period had a higher average sensor glucose, GMI, and TAR at baseline compared to those who did not attend a telemedicine visit (180 vs 159.5 mg/dl, p = 0.02; 7.7 vs 7.25%, p = 0.05; 48 vs 35%, p = 0.01; respectively). These differences are likely due to the protocol of our telemedicine clinic that prioritized patients with worse glucose control for a telemedicine visit during the lockdown period. No significant differences were noted in any of the other CGM metrics, including CGM active time and number of daily scans, at baseline between the two groups.Table 1 Baseline characteristics of study participants (n = 101). Characteristic All Study Participants (n = 101) Telemedicine visit during the lockdown period P value † No (n = 40) Yes (n = 61) Age, median (25th, 75th percentile), years 23 (18,28) 27 (19,31.5) 22 (17,26) 0.08 Gender Women (%) 54.46 47.50 59.02 0.18 Men (%) 45.54 52.50 40.98 Diabetes duration, median (25th, 75th percentile), years 7(3,16) 10 (4,17.5) 7 (3,13) 0.12 Body mass index, median (25th, 75th percentile), kg/m2 23.66 (21.36,27.66) 23.74 (21.285,27.565) 23.66 (21.36,27.66) 0.79 hemoglobin A1C, median (25th, 75th percentile), % 8.3 (7.7,9.9) 8.2 (7.5,10.6) 8.3 (7.8,9.7) 0.66 Comorbidities Hypothyroidism, n (%) 10 (11) 4 (12) 6 (10) 0.77 Celiac disease, n (%) 8 (9) 3 (10) 5 (9) 0.83 Dyslipidemia, n (%) 8 (9) 2 (6) 6 (10) 0.50 HTN, n (%) 30 (31) 7 (19) 23 (38) 0.05 Diabetic Complications Nephropathy, n (%) 10 (10) 3 (8) 7 (12) 0.59 Albuminuria, n (%) 15 (16) 4 (11) 11 (18) 0.35 Retinopathy, n (%) 7 (13) 4 (19) 3 (9) 0.29 Mode of Insulin Therapy Using MDI (%)* 71.43 70.27 72.13 0.11 Using CSII (%)* 28.57 29.73 27.87 Total Daily Dose of Insulin, median (25th, 75th percentile), Units/day 50 (37.5, 68) 44.5 (31.5, 68) 51.5 (40, 68) 0.12 CGM metrics (pre-lockdown) * Average sensor glucose, median (25th, 75th percentile), mg/dl 173 (145,204) 159.5 (138,180.5) 180 (154,208) 0.02* GMI, median (25th, 75th percentile), %* 7.5 (6.8,8.3) 7.25 (6.65,7.9) 7.7 (7,8.4) 0.05 TIR, median (25th, 75th percentile), %* 52 (35,65) 58 (45.5,73.5) 46 (34,61) 0.01* TAR, median (25th, 75th percentile), %* 43 (23,60) 35 (19.5,48.5) 48 (33,63) 0.01* TBR, median (25th, 75th percentile), %* 4 (2,7) 4.5 (2.5,8.5) 3 (2,7) 0.16 Coefficient of variation, median (25th, 75th percentile), % 37.7 (33,42.1) 37.2 (33.8,41.25) 37.8 (31.2,42.8) 0.90 CGM active time, median (25th, 75th percentile), % 92 (74,98) 93.5 (74.5,100) 92 (74,98) 0.22 Number of daily scans, median (25th, 75th percentile), n/day 10 (6,14) 10.5 (6,14) 9 (6,14) 0.39 * Abbreviations: MDI, multiple daily injection; CSII, continuous subcutaneous insulin infusion; CGM, continuous glucose monitoring; GMI, glucose management indicator; TIR, time in range; TAR, time above range; TBR, time below range. † P value examining the difference between those who attended a telemedicine visit during the lockdown period versus those who did not. 3.2 Impact of lockdown on glycemic control and CGM use Overall, the six-week lockdown period was accompanied by an improvement in average glucose (from 173 to 159 mg/dl; P = 0.01) and TAR (from 43 to 35%; P = 0.01) in people with T1D in our study (Table 2 ). When stratified by whether or not they attended a telemedicine visit during the lockdown period, people with T1D who attended a telemedicine visit had statistically significant improvements in average glucose (from 180 to 159 mg/dl; P < 0.01), GMI (from 7.7 to 7.2%; P = 0.03), TIR (from 46 to 55%; P < 0.01), and TAR (from 48 to 35%; P < 0.01) (Fig. 2 ). No significant changes were noted in any of the other CGM metrics including TBR or low glucose events. In those who did not attend a telemedicine visit during lockdown, no significant changes were noted in any of the glycemic indices on CGM (Table 2). Similarly, the frequency of CGM use remained unchanged from baseline to the end of the lockdown period in the two groups with an overall median CGM active time of 92% at baseline and 90% at the end of the lockdown period (P = 0.56) and a median number of daily scans of 10 at baseline and 10 at the end of the lockdown (P = 0.45) (Table 2). No cases of diabetic ketoacidosis (DKA) or severe hypoglycemic events have been reported in the participants’ medical records during the study period.Table 2 Changes in various CGM metrics during the lockdown period compared to those in the period before lockdown in people with diabetes who had a telemedicine visit and those who didn’t have a telemedicine visit during the lockdown period.* Variable All (n = 101) No telemedicine visit (n = 40) Had telemedicine visit (n = 61) Before lockdown During lockdown P value† Before lockdown During lockdown P value† Before lockdown During lockdown P value† Average glucose, median (25th, 75th), mg/dl 173 (145,204) 159 (137,194) 0.01 159.5 (138,180.5) 160 (140,188) 0.99 180 (154,208) 159 (135,199) <0.01 GMI, median (25th, 75th percentile), % 7.5 (6.8,8.3) 7.2 (6.6,8.4) 0.11 7.3 (6.7,7.9) 7.2 (6.7,8.0) 0.65 7.7 (7.0,8.4) 7.2 (6.6,8.5) 0.02 TIR, median (25th, 75th percentile), % 52 (35,65) 56 (38,69) 0.10 58 (45.5,73.5) 57 (43.5,71) 0.20 46 (34,61) 55 (37,69) <0.01 TAR, median (25th, 75th percentile), % 43 (23,60) 35 (22,56) 0.01 35 (19.5,48.5) 35 (22.5,51) 0.83 48 (33,63) 35 (21,57) <0.01 TBR, median (25th, 75th percentile), % 4 (2,7) 5 (2,10) 0.05 4.5 (2.5,8.5) 5.5 (2,10) 0.40 3 (2,7) 5 (2,8) 0.06 CV, median (25th, 75th percentile), % 37.7 (33,42.1) 38.9 (33.3,44.8) 0.10 37.2 (33.8,41.25) 39.4 (34.05,44.7) 0.11 37.8 (31.2,42.8) 37.7 (33,44.8) 0.33 CGM active time, median (25th, 75th percentile), % 92 (74,98) 90 (68,99) 0.56 93.5 (74.5,100) 91.5 (66.5,98.5) 0.24 92 (74,98) 90 (69,99) 0.86 Daily scans, median (25th, 75th percentile), n/day 10 (6,14) 10 (5,15) 0.45 10.5 (6,14) 8.5 (4,14.5) 0.40 9 (6,13) 11 (6,15) 0.09 Low glucose events, median (25th, 75th percentile), n/day 8 (3,14) 8 (3,14) 0.76 11 (5,16) 8 (3.5,15) 0.28 6 (2,13) 8 (3,18) 0.22 Abbreviations: GMI, glucose management indicator; TIR, time in range; TAR, time above range; TBR, time below range; CV, Coefficient of variation; CGM, continuous glucose monitoring. * Data are presented as median and interquartile ranges. † P value examining the difference in glycemic indices by period (before lockdown vs. during lockdown) using Wilcoxon matched-pairs signed-rank test in those who had a telemedicine visit and those who did not have a telemedicine visit during the lockdown period. Fig. 2 Changes in glycemic indices from pre-lockdown to end of lockdown in those who attended a telemedicine visit versus those who did not. A) Changes in Time In Range (TIR); B) Changes in Time Above Range (TAR); C) Changes in Glycemic Management Indicator (GMI). Abbreviations: TIR, time in range; TAR, time above range; GMI, glucose management indicator. * P < 0.05 comparing glycemic indices at the end-of-lockdown period to the pre-lockdown period. 4 Discussion In this study, we showed that telemedicine use, during the lockdown period in Saudi Arabia, was associated with a significant improvement in several glycemic indices in people with T1D, including an increase in TIR by 9% (i.e. 2.25 h/day) and a reduction in TAR by 13% (i.e. 3.25 h/day). These changes in TIR and TAR are both statistically and clinically significant; as they represent an approximate decrease in hemoglobin A1C of 0.8% [12], [13], [14], [15], [16], [17], [18], [19]. Each 10% increase in TIR has been associated with a reduction in risk of retinopathy by 36% and microalbuminuria by 60% in people with T1D [20]. An equivalent reduction in A1C in people with type 2 diabetes has also been associated with reduced relative risks of all-cause mortality by 14%, death due to diabetes by 21%, any end point related to diabetes by 21%, fatal and non-fatal myocardial infarction by 14%, heart failure by 16%, fatal and non-fatal stroke 12%, and microvascular endpoints by 31% over 10 years [21], [22], [23]. We also showed that lockdown was not associated with worsening nor improvement of glycemic control in the subgroup of people with T1D who did not attend a telemedicine visit, a finding that warrants further discussion. Recent reports have shown similar findings in people with T1D during lockdown in Italy and Spain [24], [25], [26], [27], [28], [29]. However, no previous study has reported the changes in glycemic control in people with T1D who attended a diabetes telemedicine clinic during lockdown versus those who did not. Moreover, the lockdown in Saudi Arabia was one of the longest and most strictly imposed lockdowns in the world as it lasted for almost 3 months and violators could face a fine of $2,665, in addition to jail for repeat offenders [4]. The lack of worsening of glucose control during lockdown in those who did not attend a telemedicine visit was likely due to several factors. It is likely that being confined to home during lockdown has eliminated factors that typically contribute to the glycemic fluctuations seen in people living with T1D, such as eating at restaurants, physical activities and exercise, and going to work or schools [24], [25], [26], [27], [28], [29]. It is also possible that people with diabetes were able to devote more time to improve their glucose control during the COVID-19 lockdown as several reports have linked poor glucose control to risk of mortality from COVID-19 [30], [31], [32]. This is supported by the remarkably high CGM active time in the two groups in our study including those who attended or did not attend a telemedicine visit during lockdown. The use of diabetes technology empowers people with T1D and has been of tremendous value during the COVID-19 outbreak and lockdown. CGM use increases the patients’ awareness of their glucose levels, compared to relying on self-monitored blood glucose, and allows them to make frequent decisions about insulin dosing and seek advice from their HCPs whenever needed. This usually results in safely achieving glycemic goals including a greater time with glucose levels at target (i.e. 70 to 180 mg/dl) and less times in hyper- and hypoglycemia. The remote glucose data sharing offered by these devices has also been key in the success of diabetes telemedicine clinics during the COVID-19 outbreak, including at our center, as it facilitates remote diabetes management, reduces patients’ exposure to COVID-19, and preserves personal protective equipment [11]. Multiple studies have reported positive perception of telemedicine in people with T1D during COVID-19 outbreak; and we have recently reported high patients’ and physicians’ satisfaction with our Diabetes Telemedicine Clinic during the COVID-19 outbreak [11], [24], [33], [34], [35]. Our study is unique as it shows the clinical effectiveness of a simplified diabetes telemedicine clinic that was rapidly implemented during the COVID-19 outbreak, utilizing technological tools that are widely available to patients and HCPs. The implications of these findings are substantial, particularly for parts of the world where diabetes is highly prevalent and telemedicine is not well-established (e.g. the Middle East and South Asia). We are not aware of similar studies examining the clinical effectiveness of diabetes telemedicine clinics during the COVID-19 outbreak. In addition, we used CGM data to assess glycemic changes during lockdown which provided reliable and detailed information that otherwise would have been difficult to obtain using self-monitoring of blood glucose. The limitations of our study include the retrospective nature of the study and lack of information on physical activity and dietary habits of the study participants during the lockdown period. In addition, our findings are limited to those who are actively using CGM and remotely sharing their data with HCPs which may impact the generalizability of our findings. 5 Conclusion Our study shows an improvement in glycemic control in people with T1D who attended a simplified diabetes telemedicine clinic, which highlights the feasibility and clinical effectiveness of this model of care during times of pandemics and disasters. Studies are needed to examine the sustainability and effectiveness of such a simplified diabetes telemedicine clinic beyond times of disasters. We also showed that the lockdown and slowing down of daily life had no immediate negative impact on glycemic control in people with T1D; however, this also needs to be examined in studies with longer duration of follow up. Disclosure The authors report no conflict of interest in this work. Acknowledgements We would like to thank the diabetes educators (Aeshah Almutairi and Eman Mohamed) and dietitians (Sara Almuammar and Nouf Alzuaibi) at the Specialized Diabetes Clinics at King Saud University-Medical City on their contribution to the Diabetes Telemedicine Clinic. Funding source This project has been supported by the College of Medicine, Deanship of Scientific Research at King Saud University, Saudi Arabia. ==== Refs References 1 Agiostratidou G, Anhalt H, Ball D, et al. Standardizing clinically meaningful outcome measures beyond HbA1c for type 1 diabetes: a consensus report of the American Association of Clinical Endocrinologists, the American Association of Diabetes Educators, the American Diabetes Association, the Endocrine Society, JDRF International, The Leona M. and Harry B. Helmsley Charitable Trust, the Pediatric Endocrine Society, and the T1D Exchange. Diabetes Care 2017;40:1622–1630 2 American Diabetes Association. 7. Diabetes Technology: Standards of Medical Care in Diabetesd2020. Diabetes Care 2020;43(Suppl. 1):S77–S88 3 Wamsley, L. Life during coronavirus: what different countries are doing to stop the spread. The Coronavirus Crisis; 2020. https://www.npr.org/sections/goatsandsoda/2020/03/10/813794446/life-during-coronavirus-what-different-countries-are-doing-to-stop-the-spread. Accessed September 4, 2020. 4 MOH News - COVID-19 Monitoring Committee Holds Its 35th Meeting https://www.moh.gov.sa/en/Ministry/MediaCenter/News/Pages/News-2020-03-25-004.aspx 5 Saudi Press Agency . Custodian of the Two Holy Mosques issues curfew order to limit spread of Novel Coronavirus from seven in the evening until six in the morning for 21 days starting in the evening of Monday 23 March. https://www.spa.gov.sa/2050402. Accessed September 4, 2020. 6 Khader M.A. Jabeen T. Namoju R. A cross sectional study reveals severe disruption in glycemic control in people with diabetes during and after lockdown in India Diabet Metabolic Syndrome Clin Res Rev 2020 7 Ghosh A. Arora B. Gupta R. Anoop S. Misra A. Effects of nationwide lockdown during COVID-19 epidemic on lifestyle and other medical issues of patients with type 2 diabetes in north India Diabet Metabolic Syndrome Clin Res Rev 2020 8 Ruiz-Roso M.B. Knott-Torcal C. Matilla-Escalante D.C. Garcimartín A. Sampedro-Nuñez M.A. Dávalos A. COVID-19 Lockdown and Changes of the Dietary Pattern and Physical Activity Habits in a Cohort of Patients with Type 2 Diabetes Mellitus Nutrients 12 8 2020 2327 32759636 9 Ghosh A. Gupta R. Misra A. Telemedicine for diabetes care in India during COVID19 pandemic and national lockdown period: guidelines for physicians Diabet Metab Syndr 14 4 2020 273 276 10 Malasanos T. Ramnitz M. Diabetes clinic at a distance: telemedicine bridges the gap Diabet Spectr 26 4 2013 226 231 11 Al-Sofiani, M. E., Alyusuf, E. Y., Alharthi, S., Alguwaihes, A. M., Al-Khalifah, R., & Alfadda, A. (2020). Rapid Implementation of a Diabetes Telemedicine Clinic During the Coronavirus Disease 2019 Outbreak: Our Protocol, Experience, and Satisfaction Reports in Saudi Arabia. Journal of diabetes science and technology, 1932296820947094. 12 Battelino T. Danne T. Bergenstal R.M. Amiel S.A. Beck R. Biester T. Clinical targets for continuous glucose monitoring data interpretation: recommendations from the international consensus on time in range Diabet Care 42 8 2019 1593 1603 13 Vigersky R.A. McMahon C. The relationship of hemoglobin A1C to time-in-range in patients with diabetes Diabet Technol Ther 21 2 2019 81 85 14 Beck R.W. Bergenstal R.M. Cheng P. Kollman C. Carlson A.L. Johnson M.L. The relationships between time in range, hyperglycemia metrics, and HbA1c J Diabet Sci Technol 13 4 2019 614 626 15 Fabris C. Heinemann L. Beck R.W. Cobelli C. Kovatchev B. Estimation of Hemoglobin A1c from Continuous Glucose Monitoring Data in Individuals with type 1 diabetes: Is Time in Range All We Need? Diabet Technol Ther ja 2020 16 Diabetes Control and Complications Trial Research Group. The effect of intensive treatment of diabetes on the development and progression of long-term complications in insulin-dependent diabetes mellitus. N Engl J Med. 1993 Sep 30;329(14):977–86. 17 Diabetes Control and Complications Trial Research Group. The relationship of glycemic exposure (HbA1c) to the risk of development and progression of retinopathy in the Diabetes Control and Complications Trial. Diabetes. 1995 Aug;44(8):968–83. 18 American Diabetes Association. 6. Glycemic targets: Standards of Medical Care in Diabetesd2020. Diabetes Care 2020; 43(Suppl. 1):S66–S76. 19 National Institute for Health and Clinical Excellence (NICE). Type 2 diabetes: newer agents for blood glucose control in type 2 diabetes; 2012. http://www. nice.org.uk/nicemedia/live/12165/44318/44318.pdf. 20 Beck R.W. Bergenstal R.M. Riddlesworth T.D. Kollman C. Li Z. Brown A.S. Validation of time in range as an outcome measure for diabetes clinical trials Diabet Care 42 3 2019 400 405 21 Selvin E. Marinopoulos S. Berkenblit G. Rami T. Brancati F.L. Powe N.R. Meta-analysis: glycosylated hemoglobin and cardiovascular disease in diabetes mellitus Ann Intern Med 141 6 2004 421 431 15381515 22 Cavero-Redondo I. Peleteiro B. Álvarez-Bueno C. Rodriguez-Artalejo F. Martínez-Vizcaíno V. Glycated haemoglobin A1c as a risk factor of cardiovascular outcomes and all-cause mortality in diabetic and non-diabetic populations: a systematic review and meta-analysis BMJ open 7 7 2017 e015949 23 Stratton I.M. Adler A.I. Neil H.A.W. Matthews D.R. Manley S.E. Cull C.A. Association of glycaemia with macrovascular and microvascular complications of type 2 diabetes (UKPDS 35): prospective observational study BMJ 321 7258 2000 405 412 10938048 24 Fernández E. Cortazar A. Bellido V. Impact of COVID-19 lockdown on glycemic control in patients with type 1 diabetes Diabetes research and clinical practice 2020 108348 25 Bonora B.M. Boscari F. Avogaro A. Bruttomesso D. Fadini G.P. Glycaemic Control Among People with Type 1 Diabetes During Lockdown for the SARS-CoV-2 Outbreak in Italy. Diabetes Therapy 1 2020 26 Maddaloni E. Coraggio L. Pieralice S. Carlone A. Pozzilli P. Buzzetti R. Effects of COVID-19 Lockdown on Glucose Control: Continuous Glucose Monitoring Data From People With Diabetes on Intensive Insulin Therapy Diabet Care 2020 27 Capaldo B. Annuzzi G. Creanza A. Giglio C. De Angelis R. Lupoli R. Blood Glucose Control During Lockdown for COVID-19: CGM Metrics in Italian Adults With Type 1 Diabetes Diabet Care 43 8 2020 e88 e89 28 Mesa, A., Viñals, C., Pueyo, I., Roca, D., Vidal, M., Giménez, M., & Conget, I., 2020. The impact of strict COVID-19 lockdown in Spain on glycemic profiles in patients with type 1 Diabetes prone to hypoglycemia using standalone continuous glucose monitoring. diabetes research and clinical practice, 108354. 29 Cotovad-Bellas L. Tejera-Pérez C. Prieto-Tenreiro A. Sánchez-Bao A. Bellido-Guerrero D. The challenge of diabetes home control in COVID-19 times: proof is in the pudding Diabet Res Clin Pract 108379 2020 30 Zhu L. She Z.G. Cheng X. Qin J.J. Zhang X.J. Cai J. Association of blood glucose control and outcomes in patients with COVID-19 and pre-existing type 2 diabetes Cell Metab 2020 31 Gupta R. Ghosh A. Singh A.K. Misra A. Clinical considerations for patients with diabetes in times of COVID-19 epidemic Diabet Metab Syndr. 14 3 2020 211 212 32 Zhou F. Yu T. Du R. Clinical course and risk factors for mortality of adult inpatients with COVID-19 in Wuhan, China: a retrospective cohort study Lancet 395 2020 1054 32171076 33 Ghosh A. Gupta R. Misra A. Telemedicine for diabetes care in India during COVID19 pandemic and national lockdown period: guidelines for physicians Diabet Metabolic Syndrome Clin Res Rev 2020 34 Odeh R. Gharaibeh L. Daher A. Kussad S. Alassaf A. Caring for a Child with Type 1 Diabetes During COVID-19 lockdown in a developing country: Challenges and Parents’ Perspectives on the Use of Telemedicine Diabet Res Clin Pract 108393 2020 35 Anjana, R. M., Pradeepa, R., Deepa, M., JEBARANI, S., Venkatesan, U., Parvathi, S. J., ... & Rani, C. S. (2020). ACCEPTABILITY AND UTILISATION OF NEWER TECHNOLOGIES AND EFFECTS ON GLYCEMIC CONTROL IN TYPE 2 DIABETES–LESSONS LEARNT FROM LOCKDOWN. Diabetes Technology and Therapeutics, (ja).

33539868

PMC9754156

NO-CC CODE

2022-12-16 23:26:28

no

Diabetes Res Clin Pract. 2021 Mar 2; 173:108682

utf-8

Diabetes Res Clin Pract

10.1016/j.diabres.2021.108682

oa_other

==== Front Comput Methods Programs Biomed Comput Methods Programs Biomed Computer Methods and Programs in Biomedicine 0169-2607 1872-7565 Elsevier B.V. S0169-2607(22)00217-6 10.1016/j.cmpb.2022.106835 106835 Article Estimating the incidence of spontaneous breathing effort of mechanically ventilated patients using a non-linear auto regressive (NARX) model Zainol Nurhidayah Mohd a Damanhuri Nor Salwa a⁎ Othman Nor Azlan a Chiew Yeong Shiong b Nor Mohd Basri Mat c Muhammad Zuraida a Chase J. Geoffrey d a Centre for Electrical Engineering Studies, Universiti Teknologi MARA, Cawangan Pulau Pinang, Permatang Pauh Campus, 13500 Pulau Pinang, Malaysia b School of Engineering, Monash University Malaysia, Bandar Sunway 47500, Malaysia c Department of Anaesthesiology and Intensive Care, Kulliyah of Medicine, International Islamic University of Malaysia, Kuantan 25200, Malaysia d Department of Mechanical Engineering, University of Canterbury, Christchurch 8041, New Zealand ⁎ Corresponding author. 26 4 2022 6 2022 26 4 2022 220 106835106835 22 11 2021 4 4 2022 21 4 2022 © 2022 Elsevier B.V. All rights reserved. 2022 Elsevier B.V. Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. Background and objective Mechanical ventilation (MV) provides breathing support for acute respiratory distress syndrome (ARDS) patients in the intensive care unit, but is difficult to optimize. Too much, or too little of pressure or volume support can cause further ventilator-induced lung injury, increasing length of MV, cost and mortality. Patient-specific respiratory mechanics can help optimize MV settings. However, model-based estimation of respiratory mechanics is less accurate when patient exhibit un-modeled spontaneous breathing (SB) efforts on top of ventilator support. This study aims to estimate and quantify SB efforts by reconstructing the unaltered passive mechanics airway pressure using NARX model. Methods Non-linear autoregressive (NARX) model is used to reconstruct missing airway pressure due to the presence of spontaneous breathing effort in mv patients. Then, the incidence of SB patients is estimated. The study uses a total of 10,000 breathing cycles collected from 10 ARDS patients from IIUM Hospital in Kuantan, Malaysia. In this study, there are 2 different ratios of training and validating methods. Firstly, the initial ratio used is 60:40 which indicates 600 breath cycles for training and remaining 400 breath cycles used for testing. Then, the ratio is varied using 70:30 ratio for training and testing data. Results and discussion The mean residual error between original airway pressure and reconstructed airway pressure is denoted as the magnitude of effort. The median and interquartile range of mean residual error for both ratio are 0.0557 [0.0230 - 0.0874] and 0.0534 [0.0219 - 0.0870] respectively for all patients. The results also show that Patient 2 has the highest percentage of SB incidence and Patient 10 with the lowest percentage of SB incidence which proved that NARX model is able to perform for both higher incidence of SB effort or when there is a lack of SB effort. Conclusion This model is able to produce the SB incidence rate based on 10% threshold. Hence, the proposed NARX model is potentially useful to estimate and identify patient-specific SB effort, which has the potential to further assist clinical decisions and optimize MV settings. Keywords Mechanical ventilation ARDS Spontaneously breathing, lung mechanics Non-linear autoregressive models (NARX) ==== Body pmc1 Introduction Acute Respiratory Distress Syndrome (ARDS) is a type of respiratory failure that causes lung inflammation, where alveoli are filled with fluid, increasing lung stiffness and surface tension resulting in the risk of lung collapse [1,2]. The Berlin ARDS definition characterizes ARDS through three levels of severity according to the ratio of the arterial partial pressure of oxygen to the fraction of inspired oxygen (PaO2/FiO2): mild ARDS: PaO2/FiO2 200–300, moderate ARDS: PaO2/FiO2 100–200, and severe ARDS: PaO2/FiO2 ≤ 100 [3]. Mortality increases with the escalation of severity and elevating at 30% - 50% [4]. Patients with ARDS are admitted to the intensive care (ICU) and require mechanical ventilation (MV) for breathing support to enable recovery [5]. The main purpose of MV in treating ARDS patients is to maintain sufficient gas exchange in the alveoli and reduces the work of breathing, while protecting the lung from additional damage. However, sub-optimal pressure and volume support in MV can cause ventilator-induced lung injury (VILI) [6]. MV is challenging due to it being highly variable and heavily relies on patient-specific response to the MV. One common strategy is to use low tidal volume and moderate driving pressures [7]. However, these values are not well defined and are also difficult to ascertain without accurate knowledge of lung mechanics. Many non-invasive model-based technique have been developed to estimate the lung mechanics throughout ventilation and aid MV management [[8], [9], [10]]. Currently, esophageal pressure measurements are used to eliminate the impact of patients’ spontaneous breathing (SB) effort on the estimated respiratory mechanics [8]. However, the application of esophageal pressure to fully represent patients’ respiratory mechanics changes. Thus, this model-based method is instigated to capture respiratory mechanics information such as lung elastance and change in pressure. The method offers an alternative in using individual patient physiology and response to MV as indicator to develop optimal ventilator settings. In particular, models and data can be used to gain insights to lung elastance and the resistance to flow. The limitation of these model-based method occurs when patients develop additional breathing effort while being sedated. Although spontaneous breathing MV modes are common and can be beneficial [11], recent supplementary studies found negative implications of SB, especially within severe form of ARDS [12]. From a model-based perspective, SB effort distorts measured airway pressure and volume measurement, thus reducing the accuracy of the model-based estimations which clearly cannot measure, or account for, SB effort [13]. In particular, SB effort reduces airway pressure, biasing estimates on lung stiffness or elastance [14]. While the electrical activity of the diaphragm can be measured, it is however highly invasive and is not considered a direct measure of the induced pressure [[15], [16]]. Equally, esophageal pressure measurement is too invasive and burdensome for regular clinical use, along with the correction method that can be too computationally intensive [[16], [17], [18], [19]]. This study aims to utilize a non-linear autoregressive (NARX) model to estimate and quantify SB efforts by reconstructing the unaltered passive mechanics airway pressure using the NARX model as suggested in [20]. The autoregressive model is able to define the input-output relationships in the system, and is capable of constructing predictions based on previous inputs [21]. Hence, the NARX model is used to reconstruct missing airway pressures caused by spontaneous breathing, where higher residual errors indicate higher incidence and greater magnitude of SB effort. Pressure will then be compared with the reconstructed airway pressure, and the mean residual error between the original and reconstructed airway pressure will be calculated. The difference between the original airway pressure and the reconstructed airway pressure is subjected to patients’ breathing and the reduction of airway pressure will result in the mean residual error to incline. Hence, this method hypothesized that higher mean residual error indicates a higher incidence of SB effort. This NARX model will be used to further investigate the characteristics of respiratory mechanics in SB patients and would potentially offer quantification and insight of the pulmonary mechanics in SB patients. 2 Method 2.1 Nonlinear autoregressive (NARX) model The nonlinear autoregressive model with exogenous inputs (NARX) is chosen as the model since the structure is represented by a simple linear difference equation. It predicts one time series given past values of the same time series, the feedback input, and another time series, called the external or exogenous time series [[21], [22]]. The model contains an error term because knowledge of other terms will not enable the current value of the time series to be predicted exactly. The model form is defined:(1) PNARX=F(Pori,t−1,Pori,t−2,Qt−1,Q,t−2)+εt where PNARX is the variable of interest which is the reconstructed airway pressure. Pori is the unaltered airway pressure and Q being the airway flow at discrete time step. In this scheme, information about airway flow, Q helps to predict airway pressure, PNARX, as do values of Pori itself. εt is the error term or sometimes called noise. Airway flow, Q, t also helps in identifying inspiration and expiration cycles based on the waveform. The function F is a pre-defined nonlinear function, such as polynomial, and can be approximated by a standard multilayer perceptron (MLP) network in this work. Fig. 1 depicts the block diagram of NARX model. The normal airway pressure, Pori is used as the output signal while airway flow, Q is used as input signal. The model computes regressors from the airway pressure and flow data, where regressors are the delayed pressure and flow. The order matrix, [na nb nk] used is [2 2 1] which defines the number of past outputs, past inputs and the input delay used in the regressor formula {Q(t-1), Q(t-2), Pori (t-1), Pori (t-2)}. The regressors are the inputs for the nonlinear and linear function blocks of the nonlinear estimator that uses wavelet network to compute predicted outputs.Fig. 1 Block diagram of NARX model. Fig 1 Below are the steps taken in applying the NARX model to reconstruct the measured airway pressure from SB patients. First, the data of airway pressure and airway flow is imported, ensuring a mix of breaths with and without SB effort as shown in Fig. 2 . All the breaths imported only focused on inspiration phase and data during expiration phase is not used. The model is trained using normal airway flow and airway pressure that are chosen manually by trained researchers from the patient's breath cycles. These normal airway flow and airway pressure are picked from breath cycles during the early ventilation period where patient least develops additional breathing. The airway flow assists in determining the breathing cycles and pattern. This data will provide sufficient information for NARX model to identify normal peak pressures.Fig. 2 The input (flow) and output (airway pressure) signals. Fig 2 In this study, there are 2 different ratios of training and validating methods, in order to choose the most optimum method as shown in Fig. 3 . The value of training data should be more than testing data to yield precise model. Firstly, the initial ratio used is 60:40 which indicates 600 breath cycles for training and 400 breath cycles used for testing. The ratio is then varied using 70:30 ratio for training and testing data. Training and estimation of airway pressure is performed using a wavelet network as the nonlinear estimator. Based on the training data set, the NARX model will calculate and derive the reconstructed airway pressure using the testing data provided for every breath. This steps are taken for the two ratios. Then, the reconstructed airway pressure (PNARX) is compared to the original airway pressure (Pori), as shown in Fig. 4 and the mean residual error is calculated. The trained model is then tested on the remaining validation data to assess performance.Fig. 3 The training data (green) and testing data (blue). Fig 3 Fig. 4 The airway pressure reconstruction using NARX. Fig 4 2.2 Patients and data The data used for this study was obtained from 10 mechanically ventilated patients treated in Intensive Care Unit (ICU) in International Islamic University Malaysia (IIUM) Medical centre [23]. The patients were fully sedated according to the ICU protocol and received Synchronised Intermittent Mandatory Ventilation (SIMV) mode (volume control) ventilation. With this mode, in SIMV, the ventilator will deliver a mandatory (set) minimum number of breaths per minute with a set volume and positive end-expiratory pressure (PEEP) while at the same time allowing spontaneous breathing. Airway pressure and flow were recorded from the ventilator at a sampling rate of 50 Hz. Table 1 tabulates the demographics of the patients enrolled in this study. A total of 10,000 breathing cycles is used in this study. The data was collected using the Clinical Application of Respiratory Elastance (CARE) software system [24]. This study is approved by the IIUM Research Ethics Committee (IREC) with approval number IRC666. The trial is registered with Australia New Zealand Clinical Trial Registry (ANZCTR) [23].Table 1 Characteristics of patients in IIUM Hospital. Table 1Patient no. Gender Age Clinical diagnostic 1 Male 54 Pneumonia 2 Male 64 Pneumonia 3 Female 72 Sepsis 4 Female 64 Pneumonia 5 Male 48 Pneumonia 6 Female 34 Pneumonia 7 Male 53 Sepsis 8 Female 61 Pneumonia 9 Male 48 Community acquired pneumonia 10 Female 53 Sepsis 2.3 Analysis Reconstructed airway pressure using NARX is compared to original, measured airway pressure in a validation set. The mean residual error is calculated based on the difference of each data point between PNARX and POri for each breath:(2) MeanResidualError=Mean(|PNARX,i−Pori,iPori,i|) Where PNARX is the reconstructed airway pressure using NARX, Pori is the original airway pressure and i is data point for every breathing cycle. This value is thus a decimal percentage deviation from the original error. Results are presented as median and interquartile range (IQR) as distributions are not Gaussian, both overall and per-patient. In this study, there are two selections of ratios are applied between training and testing data which are 60:40 and 70:30. For example, 60:40 indicates 60% of training data while 40% is the testing data. The results for both ratios are compared in order to choose which ratio of validating and training that is the most optimum. This study hypothesises high mean residual error indicates the presence of spontaneous breathing effort from the patient. Large residual error represent higher SB effort as the difference of the pressure indicates the reduction of airway pressure [25]. The residual error are evaluated further by using confusion matrix that is formed from the four outcomes produced as a result of binary classification. A binary classifier predicts all data instances of a test dataset as either positive or negative. This classification produces four outcomes – true positive (TP), true negative (TN), false positive (FP) and false negative (FN) as shown in Fig. 5 .Fig. 5 Confusion Matrix. Fig 5 The four outcomes are based on the presence of SB effort and the performance of NARX model to reconstruct pressure. In this study, TP are cases in which there are presence of SB effort and NARX model are able to reconstruct the reduced pressure. TN are when there is no SB effort and NARX model match the original pressure. FN are cases in which there are presence of SB effort but NARX model did not reconstruct the reduced pressure. Finally, FP is classified when there is no presence of SB effort but NARX reconstruct additional pressure due to model limitation. From the confusion matrix, sensitivity is evaluated using the equations below:(3) Senstivity=TPTP+FN Based on the sensitivity test, the incidence is quantified using an error threshold where these values are counted for the validation breaths per patient analysed, to create an incidence percentage. The mean residual error is also denoted as a magnitude of effort. 3 Results Fig. 6 shows the example of airway pressure and airway flow data containing both SB effort and normal breaths, where the altered pressure is clearly visible in the fifth breath while volume and flows are controlled by the ventilator. A total of 1000 breaths is assessed for each patient. These breaths are tested for two different ratio of training and testing dataset to find the best way of reconstruct the airway pressure and determine the SB effort. The NARX model used identified normal peak pressure as training data to reconstruct the missing pressure as shown in Fig. 7 . Then, the mean residual error (MRE) is calculated for every patient to validate the NARX model and is presented in median and interquartile range (IQR) value. Each example has different levels and types of SB efforts, showing the range of behaviors captured and reconstructed using the NARX model. Table 2 summarises the calculated mean residual error of all patients for all the two ratios.Fig. 6 The airway pressure (top) and flow (bottom) of Patient 1. Fig 6 Fig. 7 The airway pressure for original data and reconstructed data using NARX for both 60:40 and 70:30 ratio for (a)Patient 1 (b) Patient 2 (c) Patient 4 (d) Patient 7 (e) Patient 8 (f) Patient 10. The NARX model reconstruct the airway pressure based on the data provided for every patient. Fig 7 Table 2 Comparisons of mean residual error (MRE) for two different ratios of training and testing data for each patient. Table 2Patient No. Breathing Cycles Mean Residual Error (MSE) Med [IQR] 60:40 70:30 1 1000 0.0820[0.0660 - 0.1045] 0.0669[0.0592 – 0.0862] 2 1000 0.1211[0.0290 - 0.2475] 0.0732[0.0295 – 0.1697] 3 1000 0.0489[0.0400 - 0.0607] 0.0467[0.0388 – 0.0556] 4 1000 0.0807[0.0405 - 0.1230] 0.0651[0.0481 – 0.1262] 5 1000 0.0776[0.0149 - 0.0900] 0.0757[0.0164 – 0.0870] 6 1000 0.0165[0.0093 - 0.0330] 0.0151[0.0083 – 0.0209] 7 1000 0.0208[0.0117 - 0.0604] 0.0259[0.0147 – 0.0603] 8 1000 0.0313[0.0114 - 0.1074] 0.0343[0.0101 – 0.1270] 9 1000 0.0567 [0.0559 - 0.0577] 0.0912[0.0872 – 0.0932] 10 1000 0.0270 [0.0220 - 0.0615] 0.0215[0.0194 – 0.0471] Total 0.0557 [0.0230 - 0.0874] 0.0534 [0.0219 - 0.0870] Fig. 8 shows the comparison of PNARX versus PORI based on the threshold of MRE between 5% to 20%. The summary of the sensitivity test is tabulated in Table 3 for various threshold between 5% to 20%. It shows that 10% threshold produced the highest sensitivy test. From the sensitivity test, the incidence is quantified using an error threshold of 10% where these values are counted for the validation breaths per patient analysed, to create an incidence percentage. The cumulative distribution function (CDF) plot of all mean residual error is shown in Fig. 9 . The CDF is plotted for both ratios and the incidence of SB is quantified using an error threshold of 10%, chosen based on sensitivity test as summarised in Table 4 .Fig. 8 Comparison of PNARX vs PORI based on Mean Residual Error (MRE) sensitivity test for Patient 2 at a) MRE 5%, b)MRE 8%, c)MRE 10%, d) MRE 12% and e)MRE 20%. Fig 8 Table 3 Comparison of sensitivity for different threshold of MRE for all patients. Table 3MRE% Sensitivity (%) for all patients 5 40.58 8 53.49 10 71.83 12 37.67 20 33.52 Fig. 9 Cumulative distribution function (CDF) plot of the mean residual error for every breath cycles for all patients using (a) 60:40 ratio and (b) 70:30 ratio. Fig 9 Table 4 Incidence of SB effort based on arbitrary MSE threshold of 10% for the two different ratios of training and testing data for each patient. Table 4Patient No. Breathing Cycles Incidence of SB effort (%) 60:40 70:30 1 1000 30% 22% 2 1000 62% 44% 3 1000 6% 3% 4 1000 40% 30% 5 1000 7% 3% 6 1000 5% 1% 7 1000 5% 16% 8 1000 30% 41% 9 1000 0% 2% 10 1000 1% 0% 4 Discussion SB effort may occur even though patients are fully sedated [14], altering the measured pressure and flow waveform which masked the true underlying lung mechanics from model-based identification. These errors inhibit the ability of the model-based methods from accurately operating as guide care on patients’ ventilation settings [26,27]. In this study, a NARX model is used to reconstruct the reduced airway pressure caused by spontaneous breathing after being trained on regular airway pressure. The original data and airway pressure constructed by the NARX model are to be compared and the mean residual error is to be calculated with the intend to quantify the incidence and the magnitude of asynchronous in larger errors, where smaller errors due to noise and model identification. Ergo, when the value of the mean residual error inclines, the higher the magnitude of spontaneous breathing, and more frequently it is, the greater the incidence of SB effort [20]. This study considered each breathing cycles as independent and can be analysed separately. The analysis of a period of breathing cycles, whether the breath is big or small is unique and can be considered as normal breath. Thus, regardless of the respiratory rate, or other MV settings, every breathing cycle is considered unique and their magnitude are analysed in respect to their NARX estimated PORI. The NARX model identifies and reconstructs the airway pressure for ten patients from IIUM Hospital Kuantan, Malaysia. Two ratios of training versus testing data are selected with a focus to examine the variation in NARX predicted output, which in this case is particularly within the reconstructed airway pressure. The value of the training data should be greater than the testing data to yield precise model. The ratios applied are 60:40 which indicates 60% of training and 40% of testing to warrant the accuracy of the model and additional to the ratios of 70:30 where 70% is the training data and 30% is the testing data. As shown in Fig. 7(a), through the application of the 60:40 ratio, the NARX model is able to fit the original airway pressure of Patient 1 across every breathing cycle and is able to reconstruct the reduced pressure. The mean residual error for Patient 6 however, is lower with a median and IQR of 0.0165[0.0093 - 0.0330] denoting that the patient did not develop frequent spontaneous breathing effort during ventilation as tabulated in Table 2. In contrast, as depicted in Fig. 7, the altered pressure waveform can be seen on Patient 2 and Patient 7. The NARX model is able to reconstruct unaltered pressure waveform due to the SB effort. Considering that the difference between the original and the reconstructed airway pressure waveform is only significant at times where SB is present, the mean residual error is higher for these breaths. Hence, it is easier to monitor the incidence of spontaneous breathing based on reconstructed waveform error, where a larger error signifies a larger magnitude, and certain error threshold can be utilized to delineate the dissimilarities between asynchrony and normal breathing. Patient 2 has the highest maximum value of mean residual error with a median and IQR of 0.1211[0.0290 - 0.2475] indicating that the patient develops frequent SB effort as shown in Table 2. The mean residual error is notably lower in some breaths, such as from Patient 6 and Patient 10. The difference between the original airway pressure and the reconstructed airway pressure is relatively small. Table 2 shows that Patient 6 has the lowest mean residual error with a median and IQR of 0.0165[0.0093 - 0.0330]. Besides that, to compare the performance of the NARX model, the training and testing data are then varied using the 70:30 ratios. The sensitivity tests are evaluated on multiple values of MRE percentage which are 5%, 8%, 10%, 12% and 20%. The confusion matrix is adjusted according to the MRE percentage value. The sensitivity test is evaluated for all patients on the selected MRE threshold as shown in Fig. 8 and Table 3, where the highest sensitivity is at 10% threshold. This implies that it can capture the most incidence of SB at 10% threshold. From Fig 8, it shows that the ability of NARX to reconstruct airway pressure and estimate SB effort especially in Patient 2 at 10% threshold. Hence, SB incidence rate is observed based on the 10% threshold. The results are summarized and tabulated in Table 4. The CDF plot of mean residual error for every breath cycle for each patient is generated based on the two ratios as shown in Fig. 9. As depicted in Tables 2 and 4, Patient 2 has the highest median and IQR of residual error and has an incidence rate of 62% at 10% threshold. This implies that Patient 2 developed the highest amount of occurrence of SB effort during ventilation. In contrast, for the 70:30 ratio, Patient 2 contributes to 44% of incidence based on the 10% threshold with a median and IQR of 0.0732[0.0295 – 0.1697] which is slightly lower, in contrast to the previous result of using the 60:40 ratio. Meanwhile Patient 6 remains inert with the lowest median and IQR of 0.0151[0.0083 – 0.0209] with only 1% of the SB incidence rate. Patients with low incidence rate exhibit the act of stunting on any SB efforts. Prominently, most errors are under 5% as displayed in Table 4 and could possibly escalate either to noise/sound or other identification errors, where models are not remarkably perfect in capturing all observed dynamics [[28], [29]]. The mean residual error may also be affected by the testing data to be disparate in terms of asynchrony. It is shown in Fig. 7 between the comparison of two ratios, based on the reconstructed airway pressure shows that 70:30 ratio produced a better reconstructed model which is almost similar to the original airway pressure for most patients. This is because the NARX model which uses the 70:30 ratio has more features to predict major concrete outliers in the testing data. However, by adding more data to the training may cause a swarm of overfitting since additional features may either be irrelevant or redundant, especially when it involves noise in the pressure waveform [20,22]. Hence, this will disrupt the reconstruction of the reduced pressure. Based on the presented result, it is apparent that the NARX model is able to reconstruct the missing pressure and provides the indication of spontaneous breathing effort based on the calculation of the mean residual error, where higher mean residual error indicates greater magnitude of SB breathing efforts, and the SB incidence of 10% threshold can be counted from these occurrences. In spite of everything, one of the limitations for the model is that it requires a normal breathing waveform for training in order to identify normal peak pressure for the data to be used to reconstruct the reduced or in some cases, increase the airway pressure during SB induced asynchrony. These normal pressures are a common occurrence during the early ventilation period where patients least develop additional breathing. Therefore, it is highly suggested for future clinical work to obtain the data in the early stage when the patients are fully sedated. Nevertheless, this study focalizes on patients with the same ventilation mode, which is the volume-control mode. Therefore, the outcomes for another mode of ventilation settings may require further research to conclude its finding. Although the NARX model has this limitation, based on the results presented, this model is able to reconstruct the missing airway pressure based on the two ratios and provides the indication of spontaneous breathing effort based on the calculation of the mean residual error where higher mean residual error indicates greater magnitude of SB breathing efforts. Hence, these 10% threshold of the SB incidence will allow a new perspective in identifying the SB effort in fully sedated mechanically ventilated patients. 5 Conclusion This study presents a non-invasive NARX model that is able to reconstruct the missing airway pressure to better estimate and quantify the spontaneous breathing effort produced by the fully sedated mechanically ventilated patient. This NARX model uses simple function and potentially useful to be implemented in hospitals as it does not require additional protocols and can be used by the bedside. The SB effort is calculated based on the mean residual error which is derived based on the difference between the original airway pressure and reconstructed airway pressure. Based on the results of median and IQR for the mean residual error, the model is able to identify the spontaneous breathing effort produced by sedated patients. Furthermore, the SB incidence rate can be determined by referring to CDF plot with a 10% threshold as a reference. Hence, this new finding allows further study of the respiratory mechanics and could assist in improving the MV management and clinical decision especially for SB patients. Declaration of Competing Interest The authors declared that there is no conflict of interest. Acknowledgment Authors would like to thank Ministry of Education, Malaysia, for the Fundamental Research Funding grant (FRGS/1/2019/TK04/UITM/02/27), Universiti Teknologi MARA, Cawangan Pulau Pinang and IIUM Hospital for providing research facilities to run this study. ==== Refs References 1 Cattel F. Use of exogenous pulmonary surfactant in acute respiratory distress syndrome (ARDS): Role in SARS-CoV-2-related lung injury Respir. Physiol. Neurobiol. 288 November 2020 2021 103645 10.1016/j.resp.2021.103645 2 Kamo T. Prognostic values of the Berlin definition criteria, blood lactate level, and fibroproliferative changes on high-resolution computed tomography in ARDS patients BMC Pulm. Med. 19 1 2019 1 9 10.1186/s12890-019-0803-0 30606165 3 Kunze J. Fritsch S. Peine A. Maaßen O. Marx G. Bickenbach J. Management of ARDS: From ventilation strategies to intelligent technical support – Connecting the dots Trends Anaesth. Crit. Care 2020 10.1016/j.tacc.2020.05.005 no. xxxx 4 Liu S. Optimal mean airway pressure during high-frequency oscillatory ventilation in an experimental model of acute respiratory distress syndrome: EIT-based method Ann. Intensive Care 10 1 2020 4 11 10.1186/s13613-020-0647-z 31925610 5 Sundaresan A. Chase J.G. Shaw G.M. Chiew Y.S. Desaive T. Model-based optimal PEEP in mechanically ventilated ARDS patients in the Intensive Care Unit Biomed. Eng. Online 10 2011 1 18 10.1186/1475-925X-10-64 no. Mv 21244718 6 Sun Q. Zhou C. Chase J.G. Parameter updating of a patient-specific lung mechanics model for optimising mechanical ventilation Biomed. Signal Process. Control 60 2020 102003 10.1016/j.bspc.2020.102003 7 Morton S.E. Predictive Virtual Patient Modelling of Mechanical Ventilation: Impact of Recruitment Function Ann. Biomed. Eng. 47 7 2019 1626 1641 10.1007/s10439-019-02253-w 30927170 8 Chiew Y.S. Time-varying respiratory system elastance: A physiological model for patients who are spontaneously breathing PLoS One 10 1 2015 1 13 10.1371/journal.pone.0114847 9 Redmond D.P. Docherty P.D. Chiew Y.S. Chase J.G. A polynomial model of patient-specific breathing effort during controlled mechanical ventilation Proc. Annu. Int. Conf. IEEE Eng. Med. Biol. Soc. EMBS 2015-Novem 2015 4532 4535 10.1109/EMBC.2015.7319402 no. L 10 Arunachalam G.R. Chiew Y.S. Tan C.P. Ralib A.M. Nor M.B.M. Patient asynchrony modelling during controlled mechanical ventilation therapy Comput. Methods Programs Biomed. 183 2020 10.1016/j.cmpb.2019.105103 11 Yoshida T. Uchiyama A. Fujino Y. The role of spontaneous effort during mechanical ventilation: Normal lung versus injured lung J. Intensive Care 3 1 2015 1 7 10.1186/s40560-015-0083-6 25705426 12 Yoshida T. Uchiyama A. Matsuura N. Mashimo T. Fujino Y. Spontaneous breathing during lung-protective ventilation in an experimental acute lung injury model: High transpulmonary pressure associated with strong spontaneous breathing effort may worsen lung injury Crit. Care Med. 40 5 2012 1578 1585 10.1097/CCM.0b013e3182451c40 22430241 13 Damanhuri N.S. Negative Lung Elastance in Mechanically Ventilated Spontaneously Breathing Patient IFAC-PapersOnLine 50 1 2017 15179 15184 10.1016/j.ifacol.2017.08.2269 14 Damanhuri N.S. Assessing respiratory mechanics using pressure reconstruction method in mechanically ventilated spontaneous breathing patient Comput. Methods Programs Biomed. 130 2016 175 185 10.1016/j.cmpb.2016.03.025 27208532 15 Khirani S. On-line monitoring of lung mechanics during spontaneous breathing: a physiological study Respir. Med. 104 3 2010 463 471 10.1016/j.rmed.2009.09.014 20096552 16 Redmond D.P. Chiew Y.S. Major V. Chase J.G. Evaluation of model-based methods in estimating respiratory mechanics in the presence of variable patient effort Comput. Methods Programs Biomed. 171 2019 67 79 10.1016/j.cmpb.2016.09.011 27697371 17 Kim K.T. Knopp J. Chase J.G. Quantifying patient spontaneous breathing effort using model-based methods Biomed. Signal Process. Control 69 November 2020 2021 102809 10.1016/j.bspc.2021.102809 18 Major V. Respiratory mechanics assessment for reverse-triggered breathing cycles using pressure reconstruction Biomed. Signal Process. Control 23 2016 1 9 10.1016/j.bspc.2015.07.007 19 Kannangara D.O. Estimating the true respiratory mechanics during asynchronous pressure controlled ventilation Biomed. Signal Process. Control 30 2016 70 78 10.1016/j.bspc.2016.06.014 20 Langdon R. Docherty P.D. Chiew Y.S. Damanhuri N.S. Chase J.G. Implementation of a non-linear autoregressive model with modified Gauss-Newton parameter identification to determine pulmonary mechanics of respiratory patients that are intermittently resisting ventilator flow patterns IFAC-PapersOnLine 28 20 2015 354 359 10.1016/j.ifacol.2015.10.165 21 Langdon R. Docherty P.D. Chiew Y.S. Möller K. Chase J.G. Use of basis functions within a non-linear autoregressive model of pulmonary mechanics Biomed. Signal Process. Control 27 2016 44 50 10.1016/j.bspc.2016.01.010 22 Langdon R. Docherty P.D. Chiew Y.S. Chase J.G. Extrapolation of a non-linear autoregressive model of pulmonary mechanics Math. Biosci. 284 2017 32 39 10.1016/j.mbs.2016.08.001 27513728 23 Chiew Y.S. Clinical Application of Respiratory Elastance (CARE Trial) for Mechanically Ventilated Respiratory Failure Patients: A Model-based Study IFAC-PapersOnLine 51 27 2018 209 214 10.1016/j.ifacol.2018.11.641 24 Redmond D.P. Managing patient-specific mechanical ventilation: Clinical utilisation of respiratory elastance (CURE) - Model and software development 19 3 2014 IFAC 25 Langdon R. Modelling and Prediction of Physiological Behaviour in Critically Ill Patients 2017 University of Canterbury 26 Morton S.E. Optimising mechanical ventilation through model-based methods and automation Annu. Rev. Control 48 2019 369 382 10.1016/j.arcontrol.2019.05.001 27 Chase J.G. Next-generation, personalised, model-based critical care medicine: A state-of-the art review of in silico virtual patient models, methods, and cohorts, and how to validation them Biomed. Eng. Online 17 1 2018 1 29 10.1186/s12938-018-0455-y 29310661 28 Schranz C. Knöbel C. Möller K. Parameter identification in models of respiratory mechanics - A hierarchical approach IFMBE Proc 37 11 2011 291 294 10.1007/978-3-642-23508-5_76 29 Schranz C. Moeller K. Docherty P.D. Chiew Y.S. Chase J.G. Covariance Analysis for Practical Identifiability of an Alveolar Recruitment Model Biomed. Tech. 57 2012 47 10.1515/bmt-2012-4005 SI-1 Track-O

35512627

PMC9754157

NO-CC CODE

2022-12-16 23:26:28

no

Comput Methods Programs Biomed. 2022 Jun 26; 220:106835

utf-8

Comput Methods Programs Biomed

10.1016/j.cmpb.2022.106835

oa_other

==== Front Trends Genet Trends Genet Trends in Genetics 0168-9525 0168-9525 Elsevier Ltd. S0168-9525(21)00057-3 10.1016/j.tig.2021.03.003 Science & Society Women and Minorities Encouraged to Apply (Not Stay) Windsor Leah C. 1⁎ Crawford Kerry F. 2⁎ 1 Institute for Intelligent Systems, The University of Memphis, Memphis, TS, USA 2 Department of Political Science, James Madison University, Harrisonburg, VA, USA ⁎ Correspondence: 23 3 2021 6 2021 23 3 2021 37 6 491493 © 2021 Elsevier Ltd. All rights reserved. 2021 Elsevier Ltd Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. The Coronavirus 2019 (COVID-19) pandemic has deepened gender and racial diversity problems in academia. Mentorship shows women and other under-represented groups where the ladders to success are, and helps them avoid the chutes, a revised leaky pipeline metaphor. Here, we identify tangible strategies that will improve gender equity, including increasing active mentorship by male academics. Keywords mentorship gender COVID-19 pandemic equity ==== Body pmcThe Pandemic and the Hidden Curriculum The COVID-19 pandemic has deepened gender and racial diversity problems in academia. While many job offers encourage women and minorities to apply, the institutions do not provide the support that will retain and promote talented scholars from under-represented communities. During the pandemic, women scholars are submitting fewer articles for publication [1]. The bulk of the cognitive and emotional labor related to childcare continues to fall on women, although many men in academia appear to be experiencing ‘gender shock’ in the new work-from-home environment, where life and scholarship hang in the balance. To address these problems, many universities have instituted blanket tenure clock extensions. However, these will inevitably favor those not in caregiving roles, and exacerbate the gender gapi [2]. While academic parents are privileged in many ways to be able, in many cases, to work from home, they are not immune from the stress of balancing work and family life, and often do so in isolation. Academics rarely have the choice to live near their support systems, such as family and friends. The early years of family formation often overlap with the tenure clock. The pandemic overlaid on top of this scenario has created an untenably stressful environment, and women and mothers are bearing the bulk of the burden. Women were already tenured and promoted at lower rates compared with men, although this is improving in some disciplines more quickly than in others. A Crisis for Women in Academia The pandemic represents nothing short of a crisis for women in academia: they have greater demands on their time, and more is expected of their academic performance than it is of their male peers. In political science, researchers found no statistically significant relationship among women for the number of publications and attaining the rank of associate professor [3]; no matter how much a woman publishes, she is still at a disadvantage compared with her male peers. Furthermore, children are a significant and positive predictor of a man’s promotion but are negatively associated with a woman’s success [4]. There are many tangible action steps that can help level the playing field, and support women’s academic careers through the pandemic and beyond. Many of these actions involve demystifying the hidden curriculum: the unwritten set of professional rules that reveal systemic bias and advance some academic careers over othersii [5]. The hidden curriculum is often explained during the course of formal mentorship programs, including preparing students for a job talk, writing a compelling cover letter, and working efficiently to convert articles from idea to publication. Mentoring also addresses how to navigate the academic system: how to present work at a conference; why to attend business meetings; and how to network with other scholars. Yet, the pandemic has sidelined these in-person activities. Women scholars in fields such as political science have stepped up to provide public goods, including coordinating virtual practice job talksiii for more than 30 scholars in Fall 2020 [6], and providing regular online mentoring sessionsiv [7]. Why Mentorship Matters Mentorship helps keep women and other under-represented groups in academia. Women, especially those of color, exit academia at higher rates than do men, a phenomenon we call ‘chutes and ladders’. Mentorship provides the ladders and teaches the hidden curriculum so that women can be better prepared to avoid the chutes. In The PhD Parenthood Trap: Caught Between Work and Family in Academia, Crawford and Windsor describe the less-observable lower-order processes, often related to family formation, that result in more visible and quantifiable higher-order processes, such as gendered gaps in citations, hires, and promotions [8,9]. Results from our recent survey on mentorship during the COVID-19 pandemic show some interesting trends. For example, the percentage of survey respondents (N = 88) who have provided formal mentorship has decreased slightly, but the percentage who reported having received formal mentorship has increased (Figure 1 ). By contrast, the percentage of those who reported having provided informal mentorship has increased, while those who reported being beneficiaries of informal mentorship has decreased (Figure 2 ). These trends provide two reasons for optimism: first, even though formal mentorship programs have been interrupted by the pandemic, they are still continuing to take place; and second, the provision of informal mentorship has increased. This may mean that the conversion to remote learning has decreased some barriers to participation as people can engage virtually when they may not have been able to do so in person.Figure 1 Response to the Question: ‘Have You Provided or Received Formal Mentoring in Political Science, before and during the Coronavirus 2019 (COVID-19) Pandemic?’. Figure 1 Figure 2 Response to the Question: ‘Have You Provided or Received Informal Mentoring in Political Science, before and during the Coronavirus 2019 (COVID-19) Pandemic?’. Figure 2 Unfortunately, mentorship is still a highly gendered activity. Most formal mentorship programs are undertaken by women, for women. Our survey sample reflects this: two-thirds of the respondents identified as women. What We Can Do We can guard against women’s preventable exits from academia by making a few concerted changes. Implicit bias training should be required for faculty on hiring, tenure, and promotion committees. Women need better advocacy from their letter-writers too; recommendation letters differ systematically in gendered ways that minimize women’s achievements, and their chances of getting hired [10]. Letters for men tend to focus on quantifying their accomplishments, whereas letters for women tend to focus on social aspects, such as collegiality, and also tend to be much shorter than men’s letters of references. More women need to be nominated for awards and high-profile service roles that amplify their stature in the profession. Women tend to take care of the academic family, performing less visible service roles with a limited scope of benefit, rather than being broadly impactful. We also need to broaden our definition of what counts as professionalv [11]. The culture of the academy often requires women to present themselves as unencumbered men, such as by removing wedding rings and not disclosing pregnancies, children, or spouses during job interviews. However, women are not unencumbered men, and some men who are carrying their share of the caregiver load also experience bias. Institutions should provide a paid research sabbatical for women after the birth or adoption of each child, in addition to, not in place of, paid parental leave. In addition, we need more ‘men in the middle’, that is, those with tenure who have leverage in changing departmental and disciplinary culture to have more active mentorship and advocacy roles [12]. The Society for Political Methodology, a conference with historically greater participation by men, recently began strongly encouraging the chairs of conference panels to call on women first, because this improves overall engagement in the discussion by women in the audience [13]. This strategy also works in classroom settings at the undergraduate and graduate levels. Why It Matters Gender essentialism, where we expect women and men to behave along traditionally defined roles, is slowing the process of scientific discovery. It took decades of rejections for Katalin Karikó’s idea about mRNA to gain traction, and yet this idea pioneered the Pfizer-BioNTech COVID-19 vaccine. In international politics, it is not only societies with women leaders, but also more egalitarian and fair societies who have had fewer COVID-19 deaths [14]. This is a lesson for academia: everyone does better when women do better. Resources i https://medium.com/international-affairs-blog/snow-days-holidays-and-pandemic-quarantines-why-we-need-to-be-looking-after-parents-a3dc38413548 ii https://medium.com/international-affairs-blog/we-need-to-talk-gender-bias-and-best-practices-for-supporting-academic-parents-1d833d34f5a iii https://sites.google.com/colgate.edu/ir-virtual-practice-job-talks/home iv https://blog.oup.com/2014/11/mentorship-academic-career-political-science/ v https://medium.com/international-affairs-blog/bias-and-professionalism-c42abf65f4ba Acknowledgments We are grateful to the mentors who have supported us in our careers, and to our families who make this work meaningful. We also wish to acknowledge the labor and efforts of the many anonymous survey respondents who have helped us understand the landscape of gender, bias, and mentorship in academia. Declaration of Interests No interests are declared. ==== Refs References 1. Viglione G. Are women publishing less during the pandemic? Here’s what the data say Nature 581 2020 365 366 32433639 2. Windsor L. Crawford K.F. Snow days, holidays and pandemic quarantines: why we need to be looking after parents Medium 2020 23 Mar 3. Hesli V.L. Predicting rank attainment in political science: what else besides publications affects promotion? PS: Polit. Sci. Polit. 45 2012 475 492 4. Mason M.A. Goulden M. Do babies matter? Academe 88 2002 21 5. Crawford K.F. Windsor L.C. We need to talk: gender, bias and best practices for supporting academic parents Int. Aff. Blog 2020 14 May 6. Lupton D. IR/CP virtual practice job talks https://sites.google.com/colgate.edu/ir-virtual-practice-job-talks 7. Leeds B.A. The importance of mentoring OUPblog 2014 23 Nov 8. Crawford K.F. Windsor L. Best practices for normalizing parents in the academy: higher- and lower-order processes and women and parents’ success PS: Polit. Sci. Polit. 53 2019 275 280 9. Crawford K.F. Windsor L. The PhD Parenthood Trap: Caught Between Work and Family in Academia 2001 Georgetown University Press 10. Madera J.M. Gender and letters of recommendation for academia: agentic and communal differences J. Appl. Psychol. 94 2009 1591 1599 19916666 11. Crawford K.F. Windsor L. Bias and ‘Professionalism’ Int. Aff. Blog 2001 3 Mar 12. Windsor, L. and Thies, C. MENtorship: ‘men in the middle’ and their role as allies in addressing gender bias. PS: Polit. Sci. Polit. (in press) 13. Carter A.J. Women’s visibility in academic seminars: women ask fewer questions than men PLoS ONE 13 2018 e0202743 14. Windsor L.C. Gender in the time of COVID-19: evaluating national leadership and COVID-19 fatalities PLoS ONE 15 2020 e0244531

33771351

PMC9754211

NO-CC CODE

2022-12-16 23:26:31

no

Trends Genet. 2021 Jun 23; 37(6):491-493

utf-8

Trends Genet

10.1016/j.tig.2021.03.003

oa_other

==== Front Diabetes Res Clin Pract Diabetes Res Clin Pract Diabetes Research and Clinical Practice 0168-8227 1872-8227 Elsevier B.V. S0168-8227(21)00103-0 10.1016/j.diabres.2021.108750 108750 Article Impact of COVID-19 lockdown on glucose control of elderly people with type 2 diabetes in Italy Falcetta Pierpaolo a Aragona Michele b Ciccarone Annamaria b Bertolotto Alessandra b Campi Fabrizio b Coppelli Alberto b Dardano Angela a Giannarelli Rosa b Bianchi Cristina b Del Prato Stefano a⁎ a Department of Clinical & Experimental Medicine, Section of Metabolic Diseases & Diabetes, University of Pisa, Pisa, Italy b Section of Metabolic Diseases & Diabetes, Azienda Ospedaliero-Universitaria Pisana, Pisa, Italy ⁎ Corresponding author at: Department of Clinical & Experimental Medicine, Section of metabolic Diseases & Diabetes, Nuovo Ospedale Santa Chiara, Via Trivella, 56124 Pisa, Italy. 17 3 2021 4 2021 17 3 2021 174 108750108750 30 12 2020 8 2 2021 5 3 2021 © 2021 Elsevier B.V. All rights reserved. 2021 Elsevier B.V. Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. Aims to evaluate the effect of home confinement related to COVID-19 lockdown on metabolic control in subjects with T2DM in Italy. Methods we evaluated the metabolic profile of 304 individuals with T2DM (65% males; age 69 ± 9 years; diabetes duration 16 ± 10 years) attending our Diabetes Unit early at the end of lockdown period (June 8 to July 7, 2020) and compared it with the latest one recorded before lockdown. Results There was no significant difference in fasting plasma glucose (8.6 ± 2.1 vs 8.8 ± 2.5 mmol/L; P = 0.353) and HbA1c (7.1 ± 0.9 vs 7.1 ± 0.9%; P = 0.600) before and after lockdown. Worsening of glycaemic control (i.e., ΔHbA1c ≥ 0.5%) occurred more frequently in older patients (32.2% in > 80 years vs 21.3% in 61–80 years vs 9.3% in < 60 years; P = 0.05) and in insulin users (28.8 vs 16.5%; P = 0.012). On multivariable analysis, age > 80 years (OR 4.62; 95%CI: 1.22–16.07) and insulin therapy (OR 1.96; 95%CI: 1.10–3.50) remained independently associated to worsening in glycaemic control. Conclusions Home confinement related to COVID-19 lockdown did not exert a negative effect on glycaemic control in patients with T2DM. However, age and insulin therapy can identify patients at greatest risk of deterioration of glycaemic control. Keywords Covid-19 Lockdown Type 2 diabetes Metabolic control Age Insulin therapy ==== Body pmc1 Introduction Since its first recognition in Wuhan, China, in December 2019, the COVID-19 pandemic caused by the novel Severe Acute Respiratory Syndrome-Coronavirus-2 (SARS-CoV2) has rapidly spread across the globe. In the absence of effective treatments or vaccines, measures have been deployed to slow the spreading of the viral infection by implementing social distancing and lockdowns of large sections of the society. In Italy, a nationwide lockdown was imposed from March 9th through May 3rd, 2020. For people with diabetes the lockdown can be expected to exert a negative impact on the management of the disease due to the anxiety and depression that can be generated by the concern about the risk of infection for them and their relatives as well as because of the uncertainties about medical and pharmacologic supply and the possibility to access regularly consultation with health care providers. In spite of this view, data in people with type 1 diabetes (T1DM) have been reassuring showing no worsening of glycaemic control and in some case a modest improvement [1], [2], [3]. These subjects, however, are generally young, trained to manage their insulin therapy often on the basis of continuous/flash glucose monitoring. The population of those with type 2 diabetes (T2DM) is more heterogeneous and generally older. Interesting enough, while several reports are available for the former, little is still available for the latter, so that it is still unclear to which extent the lockdown could have impacted on diabetes management and metabolic control in individuals with T2DM [4]. To address this issue, we have evaluated changes in metabolic control before and after lockdown in a group of patients with T2DM regularly attending our outpatient diabetes clinic. 2 Materials and methods 2.1 Participants and procedures Subjects with T2DM referring to the Diabetes Clinic of our University Hospital were included in this survey only if they had no modification of anti-hyperglycaemic therapy in the 6 months before lockdown, no presence of severe systemic illness, and no treatment with drugs known to induce hyperglycaemia. Furthermore, none of the patients had SARS-CoV2 infection nor were quarantined for close contact with infected people. Anthropometric (body mass index, BMI and waist circumference, WC) and metabolic parameters (fasting plasma glucose, FPG; glycated haemoglobin, HbA1c; creatinine; estimated glomerular filtration rate, eGFR; total, LDL-, HDL-cholesterol and triglycerides) were then obtained from 304 patients with T2DM at the end of lockdown period, between June 8 to July 7, 2020. The same parameters obtained at the time of the last visit before lockdown were retrieved from electronic medical records for comparison. Biochemical determinations were performed in the central laboratory of our Hospital during the time of the study. HbA1c was measured by high-performance liquid chromatography using DCCT-aligned methods [5]. The study protocol was approved by the Ethics Committee of University of Pisa and all subjects provided voluntary consent to their data analysis. 2.2 Statistical methods Continuous variables are expressed as mean with standard deviation (SD) and median with interquartile range (IQR); categorical variables are expressed as percentages. Normality was checked using the Shapiro–Wilk test. Paired Student’s t-test was used to compare paired continuous variables with normal distribution, while the Wilcoxon Rank test was used for not-normally distributed paired variables. A uni- and multivariable logistic regression analysis was applied to evaluate the association of age, sex, BMI, diabetes duration, presence of micro- and macrovascular complications with glycaemic control potentially associated with a worsening of HbA1c defined as an increase ≥ 0.5%. Finally, a sensitivity analysis including only those subjects with the last visit within three months before lockdown was performed. Statistical significance was accepted at two-tailed P < 0.05. Data were analysed using SPSS version 25 (IBM SPSS Statistics). 3 Results Out of 1250 patients referred to the Diabetes Unit in the selected period, 946 were excluded due to change in therapy at the last visit before lockdown or because of missing HbA1c data. The main clinical characteristics of the remaining 304 patients with T2DM are shown in Table 1 .Table 1 Clinical characteristics of the entire cohort at baseline. Variable N°. Age, years 304 69.1 ± 9.2 Sex, male 304 198 (65.1) Smoking habit Never Current Former Unknown 214 99 (32.6) 31 (10.2) 84 (27.6) 90 (29.6) Hypertension 304 229 (75.3) Dyslipidemia 304 205 (67.4) CKD 304 134 (44.1) Microalbuminuria Macroalbuminuria 304 81 (26.6) 10 (3.3) DR 304 61 (20.1) DN 304 64 (21.1) CVD 304 52 (17.1) Stroke 304 10 (3.3) HF 304 5 (1.6) PAD 304 37 (12.2) AHAs 304 Lifestyle management 4 (1.3) Insulin 104 (34.2) MDI 57 (18.7) Basal 47 (15.5) Metformin 261 (85.9) Sulphonylurea 44 (14.5) DPP4i 103 (33.9) GLP1-RA 70 (23) SGLT2i 47 (15.5) Pioglitazone 20 (6.6) Acarbose 8 (2.6) Abbreviations: AHA, anti-hyperglycaemic agents; CKD, chronic kidney disease; CVD, established cardiovascular disease; DN, diabetic neuropathy; DR, diabetic retinopathy; GLP1-RAs, GLP1 receptor agonists; HF, heart failure; MDI, multiple daily injections insulin therapy; PAD, peripheral artery disease. Data are expressed as mean ± SD or frequency (%). The mean time between the pre- and post-lockdown visit was 6.5 ± 1.6 months (median 6.2 months [IQR, 5.6–7.3]). On average, pre-lockdown visit was carried out 3.1 ± 1.5 months (median 2.9 months [IQR, 2.0–4.0]) before lockdown. Table 2 shows the anthropometric and biochemical data of the whole cohort before and after lockdown.Table 2 Clinical, anthropometric, and biochemical features of T2DM patients before and after the COVID lockdown. N° Before lockdown After lockdown Mean ± SD Median (IQR) Mean ± SD Median (IQR) P BMI, kg/m2 303 29.2 ± 5 28.8 (25.7–32.4) 29.3 ± 5.2 28.7 (25.5–32.7) 0.032§ Weight, kg 303 81.5 ± 15.9 82 (71–91) 81.8 ± 16.3 82 (70–92) 0.023§ WC, cm 244 104.4 ± 12.4 103 (97–113) 105 ± 13.9 104 (97–114) 0.001§ HbA1c, % 304 7.1 ± 0.9 7 (6.4–7.6) 7.1 ± 0.9 7 (6.4–7.6) 0.600* HbA1c, mmol/mol 304 53.7 ± 10.1 53 (47–60) 54.7 ± 10.4 52.5 (47–59.7) 0.931* FPG, mmol/l 301 8.6 ± 2.1 8.3 (7.1–9.8) 8.8 ± 2.5 8.4 (7.3–9.7) 0.353* Creatinine, mg/dl 301 1 ± 0.36 0.92 (0.77–1.14) 1.1 ± 0.6 0.96 (0.79–1.23) 0.003§ eGFR, ml/min/1.73 m2 301 79 ± 23.9 80 (61–95) 76 ± 25.8 75 (58–94) 0.001* TC, mmol/l 297 4.2 ± 0.8 4.2 (3.6–4.6) 4.0 ± 0.8 3.9 (3.4–4.5) 0.021* HDL, mmol/l 297 1.3 ± 0.3 1.2 (1.0–1.4) 1.2 ± 0.3 1.2 (1.0–1.4) 0.008§ LDL, mmol/l 295 2.2 ± 0.7 2.1 (1.7–2.6) 2.1 ± 0.7 2.0 (1.5–2.5) 0.006§ TG, mmol/l 244 1.5 ± 0.9 1.3 (1.0–1.8) 1.6 ± 0.9 1.3 (1.0–1.9) 0.379§ Abbreviations: BMI, body mass index; WC, waist circumference; HbA1c, glycated haemoglobin; eGFR, estimated glomerular filtration rate; TC, total-cholesterol; HDL, high density lipoprotein cholesterol; LDL, low density lipoprotein cholesterol, TG, triglycerides. * Student’s t-test. § Wilcoxon Rank test. Overall, minor numerical changes were apparent for almost all parameters considered, though BMI, WC, and creatinine were significantly higher while eGFR, total, LDL- and HDL-cholesterol were lower after lockdown compared to baseline. No statistically different changes were found as far as FPG, HbA1c, and triglycerides are concerned. When considering only patients with last follow-up visit up to 3 months before lockdown (n = 193; age 68.5 ± 9.3 years; 68.4% male), these results were confirmed (Suppl. Table 1). Upon stratification by age, a worsening in HbA1c (defined as an increase ≥ 0.5% compared to baseline value) was more common in older patients (<60 years: 9.3%; 61–79 years: 21.3%; ≥80 years: 32.2%; P < 0.05) while there were no differences across BMI categories. Similarly, no significant differences were observed between males and females (23.6 vs 19.2%; P = 0.368). Finally, HbA1c worsening occurred more commonly among those on insulin therapy as compared to those not using insulin (28.8 vs 16.5%, p = 0.012). The effect of age and insulin therapy was fully apparent in a multivariable analysis showing that those > 80 years had 4-fold higher risk of worsening HbA1c (OR 4.62; 95% CI, 1.22–16.07) compared to those < 60 years, while the risk associated with insulin therapy was 2-fold higher (OR 1.96; 95% CI, 1.10–3.50), independently of other factors (Table 3 ). Similar associations were found in a sensitivity analysis including only individuals with last visit before lockdown within the prior 3 months (Suppl. Table 2).Table 3 Logistic regression analysis for predictors of worsened HbA1c (ΔHbA1c ≥ 0.5 mmol/mol) during lockdown. Univariate Multivariate (Backward conditional) OR 95% CI P OR 95% CI P Male sex 0.77 0.43–1.36 0.369 Age class < 60 Ref Ref Ref Ref Ref 61–80 2.64 0.90–7.74 0.077 2.36 0.79–6.99 0.121 >80 4.64 1.30–16.6 0.018 4.62 1.22–16.07 0.024 BMI class Normal w Ref Ref Over w 0.72 0.34–1.54 0.396 Obese 1.10 0.53–2.28 0.796 Microvascular Complications No Ref Ref Ref 1 1.16 0.61–2.22 0.656 2 1.93 0.90–4.12 0.089 3 1.26 0.32–4.91 0.735 Macrovascular Complications, Yes 1.36 0.70–2.26 0.363 Insulin therapy, Yes 2.05 1.17–3.61 0.013 1.96 1.10–3.50 0.022 DD, 1 year 1.00 0.97–1.03 0.903 Abbreviations: BMI, body mass index; DD, diabetes duration. Values in bold are statistically significant. 4 Discussion In the present study, we report data on the impact of the recent lockdown period related to the COVID-19 pandemic in Italy on metabolic control of individuals with T2DM, showing that minor, though statistically significant changes were detected for some parameters but not for HbA1c, despite a slight weight gain. The robustness of our data is also confirmed by the sensitivity analysis including only patients with a strict follow-up (last visit ≤ 3 months before lockdown), thus minimizing the time-dependency of the results here reported. Our results are at variance with those reported by Khare et al. in a study involving 143 patients with T2DM in whom glycaemic control, as determined on self-monitoring, worsened during the first 3 weeks of lockdown mainly because of higher post-prandial glucose levels [6]. The authors interpreted those results as the effect of changes in diet and less physical activity occurred during the lockdown. On the contrary, Anjana et al. in a survey including 205 patients with T2DM found a significant improvement in HbA1c after lockdown (7.7 ± 1.7 vs 8.2 ± 1.9%, P < 0.001) [7]. More recently, in a series of 114 individuals with T2DM, Biancalana et al. reported no significant change in glucose control, although a 0.3% increase in HbA1c was found in 26% of them [4]. In summary, a certain degree of heterogeneity has been found as far as changes in glycaemic control are concerned in people with T2DM throughout the lockdown imposed to prevent the spreading of Sars-Cov-2 pandemic. Several reasons may contribute to such heterogeneous results, including differences in ethnicity, baseline glycaemic control and access to diabetes consultation during lockdown. Baseline HbA1c value in the study by Anjana et al. was higher compared to that of our population (8.2 vs 7.1%). Furthermore, our patients may not reflect a more general diabetic population as all of them regularly attended a tertiary care Diabetes Unit that continued providing teleconsultation during the lockdown period. Although overall no changes were detected in glycaemic control, a closer look revealed that glucose deterioration could occur in some subgroups. Thus, the percentage of the patients who had, over the lockdown, an increase of HbA1c ≥ 0.5% was greater among the elderly and those on insulin therapy. These two parameters, age > 80 years and insulin therapy, were independently associated with significant glycaemic worsening in a multivariable analysis and, as such, they could help identifying subjects for whom it may be necessary to ensure sufficient contact and surveillance during challenging time as it was the case in the lockdown and as it has been suggested in a recent survey by Bonora et al. [8]. These authors compared accesses to the diabetes centre before and during lockdown to suggest that are the elderly patients with T2DM, i.e. those with more sever burden of complications and often requiring more complex treatment, who are likely to encounter more difficulties in stay in touch with their diabetes clinics. For these people it may be more difficult to get acquainted to telematic visit and monitoring systems due to poorer familiarity with modern technologies. Insulin use also was an independent predictor associated with 2-fold higher odds of glycaemic worsening compared with use of other glucose lowering agents. This may well reflect the increased complexity of the management of this therapeutic approach, particularly for those with T2DM, since evidence currently available for patients with T1DM on continue glucose monitoring show that glycaemic control did not worsen or even improved during lockdown [1], [9], [10], [11], [12]. The latter, however, are younger, on continuous or flash glucose monitoring and more intensively instructed how to handle multiple dose insulin therapy or even continuous subcutaneous glucose infusion. Although ours as well as other results so far available may suggest a limited impact of the lockdown on metabolic control of people with T2DM, the duration of the lockdown may have been too short to fully appreciate what could be the impact of a relaxation of diabetes management that may occur under such circumstance. In line with this caution is the modest yet statistically significant increase in body weight and waist circumference that may well reflect the initiation of a trajectory that may lead to more substantial weight gain and, ultimately, deterioration of glycaemic control. Recently published surveys showed that roughly 22% of people reported gaining weight during self-quarantine along with reduced physical activity and worse eating behaviours during the COVID-19 lockdown [13], [14]. Unfortunately, due to the retrospective design of the study, data about the change in daily diet and physical activity during lockdown were not available. Nevertheless, since our patients displayed an overall stable glycaemic control, we may assume that the effect of lifestyle modifications during lockdown was negligible. Some limitation of our study needs to be acknowledged. This includes the relatively small number of participants, although ours is the largest cohort of T2DM so far reported. Also, as already pointed out, we have recruited patients regularly attending a specialized diabetic clinic thus limiting the generalizability of our results to a broader diabetic population. Finally, the duration of the lockdown may not be sufficiently long to allow a more careful assessment of the potential impact of longer lockdown and its psychological and logistic implications. In conclusion, the home confinement related to the COVID-19 lockdown, at least with the duration our patients have been exposed to, doesn’t seem to have exerted a negative effect on glycaemic control of patients with T2DM, despite slight weight gain. Nonetheless, some clinical features, in particular advanced age and insulin therapy, seem to be identify subgroups of patients with greater risk of glucose control deterioration. These characteristics may help in addressing patients requiring more attention - if not special protection - by developing special programmes at the time of challenging societal situations. Funding This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors. Appendix A Supplementary material The following are the Supplementary data to this article:Supplementary data 1 Supplementary data 2 Acknowledgments We would like to express our gratitude to all the patients attending the Diabetes Unit of Azienda Ospedaliero Universitaria Pisana. Appendix A Supplementary data to this article can be found online at https://doi.org/10.1016/j.diabres.2021.108750. ==== Refs References 1 Bonora B.M. Boscari F. Avogaro A. Bruttomesso D. Fadini G.P. Glycaemic Control Among People with Type 1 Diabetes During Lockdown for the SARS-CoV-2 Outbreak in Italy Diabetes Ther 11 2020 1 11 2 Verma A. Rajput R. Verma S. Balania V.K.B. Jangra B. Impact of lockdown in COVID 19 on glycemic control in patients with type 1 Diabetes Mellitus Diabetes Metab Syndr 14 2020 1213 1216 32679527 3 Fernández E. Cortazar A. Bellido V. Impact of COVID-19 lockdown on glycemic control in patients with type 1 diabetes Diabetes Res Clin Pract 166 2020 108348 4 Biancalana E. Parolini F. Mengozzi A. Solini A. Short-term impact of COVID-19 lockdown on metabolic control of patients with well-controlled type 2 diabetes: a single-centre observational study Acta Diabetol 2020 10.1007/s00592-020-01637-y 5 Mosca A. Goodall I. Hoshino T. Global standardization of glycated hemoglobin measurement: The position of the IFCC Working Group Clin Chem Lab Med 45 2007 1077 1080 17867998 6 Khare J. Jindal S. Observational study on Effect of Lock Down due to COVID 19 on glycemic control in patients with Diabetes: Experience from Central India Diabetes Metab Syndr 14 2020 1571 1574 7 Anjana R.M. Pradeepa R. Deepa M. Acceptability and Utilization of Newer Technologies and Effects on Glycemic Control in Type 2 Diabetes: Lessons Learned from Lockdown Diabetes Technol Ther 22 2020 527 534 32522031 8 Bonora B.M. Morieri M.L. Avogaro A. Fadini G.P. The Toll of Lockdown Against COVID-19 on Diabetes Outpatient Care: Analysis From an Outbreak Area in Northeast Italy Diabetes Care. 44 2021 e18 e21 33127611 9 Tornese G. Ceconi V. Monasta L. Carletti C. Faleschini E. Barbi E. Glycemic Control in Type 1 Diabetes Mellitus During COVID-19 Quarantine and the Role of In-Home Physical Activity Diabetes Technol Ther 22 2020 462 467 32421355 10 Christoforidis A. Kavoura E. Nemtsa A. Pappa K. Dimitriadou M. Coronavirus lockdown effect on type 1 diabetes management οn children wearing insulin pump equipped with continuous glucose monitoring system Diabetes Res Clin Pract. 166 2020 108307 11 Maddaloni E. Coraggio L. Pieralice S. Carlone A. Pozzilli P. Buzzetti R. Effects of covid-19 lockdown on glucose control: Continuous glucose monitoring data from people with diabetes on intensive insulin therapy Diabetes Care 43 2020 e86 e87 32503838 12 Aragona M. Rodia C. Bertolotto A. Type 1 diabetes and COVID-19: The “lockdown effect” Diabetes Res Clin Pract 170 2020 108468 13 Zachary Z. Brianna F. Brianna L. Self-quarantine and weight gain related risk factors during the COVID-19 pandemic Obes Res Clin Pract 14 3 2020 210 216 32460966 14 Ammar A. Chtourou H. Boukhris O. COVID-19 Home Confinement Negatively Impacts Social Participation and Life Satisfaction: A Worldwide Multicenter Study Int J Environ Res Public Health 17 2020 6237 10.3390/ijerph17176237 32867287

33722703

PMC9754212

NO-CC CODE

2022-12-16 23:26:31

no

Diabetes Res Clin Pract. 2021 Apr 17; 174:108750

utf-8

Diabetes Res Clin Pract

10.1016/j.diabres.2021.108750

oa_other

==== Front J Econ Behav Organ J Econ Behav Organ Journal of Economic Behavior & Organization 0167-2681 0167-2681 Elsevier B.V. S0167-2681(20)30451-0 10.1016/j.jebo.2020.12.004 Article Online cheating amid COVID-19☆ Bilen Eren a Matros Alexander ab⁎ a Darla Moore School of Business, University of South Carolina, Columbia, SC 29208, United States b Lancaster University Management School, Lancaster LA1 4YX, United Kingdom ⁎ Corresponding author. 28 12 2020 2 2021 28 12 2020 182 196211 23 10 2020 3 12 2020 © 2020 Elsevier B.V. All rights reserved. 2020 Elsevier B.V. Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. We present evidence of cheating that took place in online examinations during COVID-19 lockdowns and propose two solutions with and without a camera for the cheating problem based on the experience accumulated by online chess communities over the past two decades. The best implementable solution is a uniform online exam policy where a camera capturing each students computer screen and room is a requirement. We recommend avoiding grading on a curve and giving students less time but simpler questions on tests. Keywords Education Cheating Online examinations Online chess ==== Body pmc1 Introduction “No matter what the game is, when there are benefits from winning, you have cheating.” - Arkady Dvorkovich, FIDE President, 2020.⁎⁎ The COVID-19 pandemic changed the lives of all people globally with most activities being forced to move online, including teaching. Most schools and universities moved from face-to-face to online delivery in March 2020. Among other difficulties related to online teaching, measuring student performance became one of the chief concerns of instructors. Many universities reported widespread cheating in online examinations that took place in Spring 2020, and the problem became so rampant that even the media addressed it. See, for example, two recent articles in Washington Post (Newton 2020) and Inside Higher Ed (Lederman 2020).1 The 2020 Advanced Placement (AP) examinations illustrate the difficulty of measuring true student performance on online examinations without proctoring. Fig. 1 shows surges of Google searches on keywords related to exam topics perfectly correlating with the time of the examinations. Since the online environments used for the 2020 AP exams had no proctoring, many students took advantage of having immediate access to Google search.2 Fig. 1 Google search trends around the time of the 2020 AP Exams. Note: 2020 AP Exams were held online due to COVID-19 related school closures. Hourly online search data is obtained from Google Trends. The search data covers the U.S. nationwide. Fig. 1 Currently, most schools and universities are occupied with the switch to online teaching. Consequently, will the cheating problem in the fall be significant enough for schools and universities to take strict measures for the future? Will it be possible to have a fair assessment system if schools and universities decide to take no action? How can schools and universities maintain academic integrity in online examinations? We must wait several months to get clear answers. However, we can make predictions on the possible outcomes by considering theory and past evidence on cheating. The problem of cheating in online environments is not new. Online chess, in particular, has been plagued by cheating ever since chess was first introduced to the internet, with players gaining an unfair advantage by using computer assistance.3 Online chess started when the Internet Chess Club (ICC) was established in 1995. The ICC first ran annual Dos Hermanas online blitz tournaments with monetary prizes in the early 2000s. Games were not proctored, and a whole array of cheating scandals consequently arose, with many ways to cheat in those events. In order to function, the ICC had to disqualify numerous people, including a former Junior World Champion, a top Chinese player, a top German player, hundreds of titled players, and thousands of amateurs (who enjoyed beating titled players). The main way to cheat was to use chess computer programs that found the best moves. Now, Chess.com is the most popular online chess club with many tournaments including monetary prizes. Unsurprisingly, cheating has surfaced as a huge problem, prompting Chess.com to create its own cheating detection unit. See Chess.com Fair Play And Cheat-Detection.4 The website states: “Though legal and practical considerations prevent Chess.com from revealing the full set of data, metrics, and tracking used to evaluate games in our fair-play tool, we can say that at the core of Chess.com’s system is a statistical model that evaluates the probability of a human player matching an engine’s top choices, and surpassing the confirmed clean play of some of the greatest chess players in history.” Cheating before the pandemic on both the ICC and Chess.com is similar to the online cheating problem that arose in Spring 2020. Thriving throughout COVID-19, cheating skyrocketed for online chess as well. For example, Chess.com announced on August 19, 2020 that it closes more than 500 accounts every day for cheating and has closed over 400,000 accounts in total, projecting to close 1,000,000 accounts by mid-2023. Of those closed accounts, nearly 400 were titled players. The only seeming positive statistic that was found indicated that female players cheat much less, only accounting for 4.57% of all title players.5 However, recent weeks revealed an explosion of top women players cheating in both online and over the board tournaments as well. Former Women’s World Champion, Anna Ushenina, was accused of cheating after her Internet 2020 Grand Prix victory. Another example is Patrycja Waszczuk, a titled young chess player, medalist of the Polish Championships, and medalist of the European Chess Championships, who was banned online and also caught cheating during her game at the Chess Festival in Ustron.6 If we can learn anything from online chess, then the message is very clear: online cheating will only get much worse and schools and universities will have their first hand experience already in Fall 2020. While online chess websites are private ventures and can ban any player for any reason, schools and universities will have a much more difficult task to provide clear evidence that proves students’ cheating. Both the ICC and Chess.com have been successful to some degree in dealing with the cheating problem although it is nowhere near to being solved. Interestingly, there are similarities in addressing the problem by both chess websites. First, they do not reveal their detection systems. Second, their disqualification decision is final. This approach admits that the detection system is vulnerable to knowledgeable cheaters. Since the websites do not have the resources to check millions of games, they implement a simple way to address the problem: a chess website monitors particular characteristics of play, and its methods of analyzing these characteristics are not revealed to the players. Players do not know what the website is looking for, making cheating more difficult.7 This cheating behavior supports the mounting evidence that the lack of “perfect honesty” exists in situations where the returns to dishonesty are high. Numerous studies using different settings and samples investigated in Gneezy (2005), Fischbacher and Föllmi-Heusi (2013), Gächter and Schulz (2016), and Vanberg (2017) show that people are more likely to deceive if the marginal benefit from deception is significantly large. Therefore, professional competitions and examinations have to use extensive monitoring to prevent cheating. However, lockdowns due to the COVID-19 pandemic in early 2020 made monitoring very difficult (or impossible) in many situations. Online tests are done without face-to-face proctoring, which implies that students can potentially use their notes, internet search, and any other assistance to help them solve questions. Furthermore, they can communicate via teleconference (or some other method) and collaborate during their exams. This cheating behavior on online examinations imposes a negative externality on students who do not cheat, especially if the instructor curves the exam scores. In this paper, we first consider two simple models of face-to-face and online examinations. These models suggest that unlike the face-to-face examination, cheating should be expected in the online examination, with the reason being very intuitive: the instructor can observe cheating evidence in the face-to-face examination, but there is only indirect cheating evidence in the online examination. Therefore, cheating is a part of the student equilibrium strategy in the online examination. We then present evidence of cheating that took place in an online examination held as part of a course taught at a large public university in Spring 2020 during COVID-19 lockdowns. Using a simple way to detect cheating - timestamps from the students’ Access Logs - we identify cases where students were able to type in their answers under thirty seconds per question. We found that the solution keys for the exam were distributed online, and these students typed in the correct as well as incorrect answers using the solution keys they had at hand. Now we suggest how to mitigate cheating based on the experience accumulated by online chess communities in the last two decades. Currently, many universities are requiring students to purchase and use a camera to record themselves while taking an exam in order to crack down on cheating, but these rules conflict with privacy rights from some students perspectives. In order to address this issue based on our theoretical models, we suggest that instructors present their students with two options: (1) If a student voluntarily agrees to use a camera to record themselves while taking an exam, this record can be used as evidence of innocence if the student is accused of cheating; (2) If the student refuses to use a camera due to privacy concerns, the instructor should be allowed to make the final decision on whether or not the student is guilty of cheating, with evidence of cheating remaining private to the instructor. Both options are designed to “implement” the outcome of the face-to-face examination when cheating is not expected in the equilibrium. Of course, there are other ways to achieve the same outcome. For example, students can take exams at proctoring centers. The implications of this paper are simple: if no action is taken for online exams in the upcoming semester, there will be widespread cheating. Students have much to gain while the probability of being caught with definitive evidence is close to zero. Using online proctoring services involving the use of a camera is one solution - albeit imperfect - to the problem of cheating. We believe that cheating can never be fully detected online and therefore recommend that instructors stay away from curving their grades in order to not punish honest students. The rest of the paper is organized as follows: Section 2 presents mixed evidence about online and face-to-face cheating, Section 3 provides two theoretical models for face-to-face and online exams, Section 4 presents examples of online cheating, and Section 5 concludes. 2 Related literature Educational institutions have traditionally been using proctoring in order to ensure academic integrity on examinations. Online education, however, typically relies on unproctored examinations that are held online. Previous surveys exploring cheating in online examinations generally claim that there is little to no difference between face-to-face and online examinations in terms of cheating. Watson and Sottile (2010) surveyed 635 students from a medium-sized university and asked whether or not they had previously cheated on an examination. They found that 32.1% of students from face-to-face courses admitted to cheating. For students from online courses, the admitted cheating rate was 32.7%. Observing that these rates are very similar, they claim that online courses do not involve more cheating. However, the main concern regarding their methodology is that they rely on self-reporting which requires students admitting they have cheated rather than actually using a mechanism to detect cheating. The next set of research addressing cheating concerns in online education includes Fask et al. (2014) which used random assignment of students to face-to-face and online examinations. They first assessed student performance using practice tests and found that the online test-takers scored 14% lower than those who took the practice test held proctored in class. However on the actual test, online test-takers scored 10% higher than the face-to-face test-takers. While their methodology had limitations in terms of detecting cheating, it provides suggestive evidence on cheating for students who take unproctored online examinations. More concrete evidence on cheating in online environments is presented in Dee and Jacob (2012), Karim et al. (2014), and Diedenhofen and Musch (2017). Using a text-based algorithm that detects plagiarism, Dee and Jacob (2012) show that 112 out of 1200 papers submitted on the Blackboard from a sample of 28 universities were plagiarized. They suggest that giving a quick tutorial explaining how plagiarism jeopardizes academic integrity at the beginning of the semester is an effective tool in preventing plagiarism. However, this may not be as effective for more direct cases of cheating. Using evidence from laboratory and online experiments, Diedenhofen and Musch (2017) show that participants cheat more (via Google search) when monetary incentives are higher. They use a computer program that triggers a pop-up message if a participant frequently changes browser tabs in a short period of time with the message stating that the researchers are aware of the participant’s cheating activity, which reduces cheating sharply. In another experimental setting using participants from Amazon’s Mechanical Turk, Karim et al. (2014) show that low-cost technology, such as web-cameras, are effective at decreasing cheating without necessarily impacting test performance. However, they observe negative reactions from a portion of the participants pointing out that these web-cameras may be viewed as invasive and thus raise feelings of pressure and tension. There is vast literature exploring cheating and deception. Becker (1968) was the first to provide the rationale for individuals who take part in illegal activities. An empirical investigation on cheaters and their incentives in Duggan and Levitt (2002) show that individuals are indeed more likely to cheat if they view returns to cheating are high and that these returns come with little cost.8 Field experiments using different settings reveal that individuals deceive more if the cost of deception is small; see Gneezy (2005), Erat and Gneezy (2012), Gächter and Schulz (2016), Vanberg (2017), Martinelli et al. (2018), Charness et al. (2019), Alan et al. (2020), and Maggioni and Rossignoli (2020). 2.1 Sequential-move game In the sequential-move game, the student chooses to either cheat or be honest. The professor observes the student choice and decides either to report the student for cheating or not. There are four outcomes in this game, but the professor and the student rank these outcomes differently. The professors outcomes from the best to the worst are (honest, do not report), (cheat, report), (cheat, do not report), (honest, report), where we record paths of play in brackets. The students outcomes from the best to the worst are (cheat, do not report), (honest, do not report), (honest, report), (cheat, report). See Fig. 2 .Fig. 2 Game Tree for the sequential move game. Fig. 2 It is easy to find a unique subgame perfect equilibrium outcome, where the student is honest and the professor does not report the student. Note that this is the best outcome for the professor and the second best outcome for the student. This sequential-move game is supposed to be played between the student and the professor during in-class exams. In the current situation, one way to implement this game is to use a camera to record the student while taking an exam. However, many students say that camera use conflicts with privacy rights and advocate taking exams without cameras. In other words, these students suggest to play the following simultaneous-move game. 2.2 Simultaneous-move game Let us consider a simple simultaneous-move game between a student and a professor. The student has two actions: cheat or be honest, and the professor also has two actions: report the student for cheating or not.9 There are four outcomes in this game, and the professor and the student rank these outcomes differently. The professor’s outcomes from the best to the worst are (honest, do not report), (cheat, report), (cheat, do not report), (honest, report). The student’s outcomes from the best to the worst are (cheat, do not report), (honest, do not report), (honest, report), (cheat, report). Table 1 gives an example of players’ payoffs. This game has a unique mixed-strategy equilibrium, which means that the student and the professor should randomize between their two actions in equilibrium. Thus cheating as well as reporting is a part of the equilibrium.Table 1 Payoff Matrix for the simultaneous move game. Table 1 Professor Report Not Student Cheat 1,3 4,2 Honest 2,1 3,4 In the mixed-strategy equilibrium, the student randomizes between her two choices in such a way to make the professor indifferent between his two choices. So, in order to determine the equilibrium probability of the student’s cheating, we have to look at the professor’s payoffs. To make our point clear, we restrict our attention on a simplified payoff Table 2 , where we only record the professor’s payoffs. Moreover, we normalize the best payoff at one and the worst payoff at zero, and 0≤c≤b≤1.Table 2 Payoff Matrix for the Professor. Table 2 Professor Report Not Student Cheat .,b .,c Honest .,0 .,1 It is easy to find now that the equilibrium probability of the student cheating, p, is equal to(1) p=11+(b−c). If the professor does not feel a big difference between (cheat, report) and (cheat, do not report) outcomes, or the difference (b−c) is small and close to zero, then the cheating probability is the highest, and in the extreme case, if (b−c)=0, this probability is equal to one, p=1. However, if the professor is concerned and sees a significant difference between (cheat, report) and (cheat, do not report) outcomes, then the student cheating probability goes down. 2.3 The problem and the solution Note that the student is honest in the unique subgame perfect equilibrium in the sequential-move game, which approximates in-class exams. However, the student is supposed to cheat in the unique mixed-strategy equilibrium in the simultaneous-move game, which is a proxy for the online exams. These findings demonstrate that cheating should be higher in online tests, and these observations are intuitive, with many instructors having first hand experience with them from face-to-face and online teaching.10 The main problem revolving around online tests is how to prove cheating. Typically, students require to see evidence of their cheating and the professor only has indirect evidence of cheating. Having only indirect evidence makes it much harder to prove that cheating took place in case of an academic integrity referral. Thus, the professor is reluctant to report cheating in online exams, or the difference b−c is close to zero in expression (1) for the equilibrium cheating probability. This in turn encourages even more cheating.11 How can this evidence problem be resolved? Many instructors have their own statistical evidence of students cheating. Some of these statistics are simple but very efficient. We present one such statistic – time spent per question – in the next section. However, once such a statistic is revealed to students, the instructor would not be able to use it again because students adjust accordingly. The solution is not to reveal information based on which the student was found guilty of cheating. In other words, if the professor claims that the student was cheating and this decision is final, then we indeed get the simultaneous-move game from the previous section. We suggest to offer two options to a student. If the student buys a camera and uses it during the exam, then the sequential-move game is played, cheating is not expected (in the equilibrium), and both the student and the professor have evidence of the student’s behavior on the exam. Alternatively, the student can have an exam without a camera in the privacy of their own home. In this case the simultaneous-move game is played, some cheating is expected in the equilibrium, and if the professor has evidence of cheating and claims cheating, then the student cannot request any evidence and appeal the verdict. 3 Data: evidence Finding concrete evidence on cheating in an online examination is potentially challenging, as demonstrated in earlier studies (Watson, Sottile, 2010, Fask, Englander, Wang, 2014). In this section, we present cases in which students were able to correctly solve several questions under thirty seconds per question.12 The mechanism we use to identify cheating is the “Access Log” provided on the Blackboard. The Access Log provides detailed timestamps which show exactly how much time a student spends on each question. Many students appear to be unaware that the time they spend on each question is recorded although they seem to expect that the information on the “total time” they take for the exam is recorded.13 Our data comes from students who were enrolled in an intermediate-level course in Spring 2020 at a large public university. The course had three Midterms and had an optional Final Exam which replaced the lowest Midterm exam. The first two Midterms were held face-to-face with proctoring; the third Midterm and the Final Exam were held online asynchronously on the Blackboard following COVID-19 related campus closures. On these online exams, students received the same set of questions in a random order. The questions were all short answer questions: the student had to type in the correct answer to receive credit with no multiple choice options given. In order to move to the next question, the student had to save and submit their answer; no moving back and forward was allowed. Table 3 presents scoresheets for two particular students who took the course in Spring 2020. Figs. 3 –4 show how much time each student spent on each question during Midterm 3 and the Final Exam with responses in Tables 4 –5 . On Midterm 3, students have both correct and incorrect answers and had to spend some time reading the problems and working to solve them.Table 3 Student 1 and 2’s scoresheets. Table 3Exam Score Letter (a) Student 1’s scoresheet Midterm 1 35/100 F Midterm 2 55/100 F Midterm 3 30/100 F Final 95/100 A (b) Student 2’s scoresheet Midterm 1 50/100 F Midterm 2 40/100 F Midterm 3 50/100 F Final 95/100 A Fig. 3 Time spent per question for Student 1. Note: Student’s Midterm 3 score is 6/20; Final score is 19/20. Total time he spent on Midterm 3 was 1 h and 10 min. Total time he spent on the Final exam was 11 min. Fig. 3 Fig. 4 Time spent per question for Student 2. Note: Student’s Midterm 3 score is 10/20; Final score is 19/20. Total time he spent on Midterm 3 was 1 h 20 min. Total time he spent on the Final exam (excluding Question #7) was 36 min. Fig. 4 Table 4 Student 1's answers on Midterm 3 (left) and Final (right). Table 4 Table 5 Student2′s answers on Midterm 3 (left) and Final (right). Table 5 Their time allocation, combined with their performance, shows no strange results for Midterm 3. However, their Access Logs reveal very peculiar information for the Final Exam. Figs. 3 and 4 reveal cases where students had to spend less than thirty seconds to solve a question. However, each of these questions requires complex problem-solving skills, demonstrating that a student would need to spend a reasonable amount of time to find each solution. Furthermore, the students had to type in their answers since the exam was not a multiple choice exam. The questions typically had non-trivial answers such as “6534” or “650” which would make it very challenging to randomly guess the correct answers. In fact, the probability of randomly guessing the correct answers on this exam is much less than the probability of winning the lottery.14 Furthermore, there is evidence that these two cases are connected. Students’ answers for all twenty questions on the Final Exam perfectly match. Both students made only one mistake on the same question where they both submitted the same incorrect answer of “125”. Fig. 5 shows the answers submitted by the rest of the class on this particular question, and it appears only three students submitted “125” while the rest of the class submitted a whole range of different numbers. Two of these three students are Students 1 and 2.15 Lastly, timestamps from their Access Logs show that once Student 2 finished his exam, Student 1 started his immediately (in 2 min) after Student 2 finished submitting his answers. We believe the probability that these students cheated and cooperated is higher than a random statistical occurrence.Fig. 5 All responses for the question both Students 1 & 2 made their only mistake on Note: Both Students 1 & 2 gave the same incorrect answer of “125”. This question was by far the most difficult question on the exam with a correct response rate of only 19.3%. The third student who submitted “125” scored 18/20 on the exam. His other incorrect answer was “12”, and the correct answer for that question was “124”. It appears he simply must have made a typo. Fig. 5 How did these students cheat? The most likely explanation is that they used online resources, where private tutors helped them solve problems. In fact, we have found evidence that the answer key for the Final Exam was distributed online in a common web platform. For a price of several dollars, students could get access to the solution key. Once the student obtained the solution key, the only task they would need to complete would be to type in the answers for the questions they were presented in a random order.16 Why did these students cheat? They had a lot to gain. Had they not cheated, they were very likely to fail the class. Student 1 had accumulated an overall score of 47.8/100 and Student 2 had 53.5/100 before the Final Exam. Their only chance to pass the class (getting at least 70) was to perform extraordinarily well on the Final Exam. Having received 95/100 each on the Final Exam, both students would pass the class. The students’ relative performances on Fig. 6 and Figs. 10 and 11 show their extraordinary performance on the Final Exam relative to the rest of the class. Furthermore, these students performed even more remarkably compared to students who took the very same course within the past 10 years. Fig. 7 shows how students - who accumulated a failing score up to the final exam - performed on the final exams given since 2010. Out of 68 such students, only 4 managed to secure a high enough score to attain a passing letter grade. These students are Students 1 and 2, as well as two other students from the same section in 2020.Fig. 6 Student 1 &2′s performance relative to the rest of the class. Note: Midterms 1 and 2 were held face-to-face with proctoring; Midterm 3 and the Final Exam were held online asynchronously without proctoring following COVID-19 related campus closures. Each dot represents a student’s test score. Fig. 6 Fig. 7 Score distribution on all final exams given between 2010–2020 for students with a score <60 before the final exam. Note: Total # of students across 10 years with a score <60 up to final exam and manage to pass the course with a C is only 4. These students are Student 1, Student 2, Student 3 (and the fourth student from the same section.) Student 3 gave identical answer keys (with 1 exception) with Students 1 and 2. Fig. 7 Note that there is nothing that stops students to type in their answers “slowly” which would mimic a case with no cheating. It appears these two students were not aware that their Access Logs had timestamps showing how much time they spent on each question. Had they known, they would have most likely submitted their answers in a longer time period so that their Access Log would look perfectly “normal”. Thus it is essential that any information on the “cheating-detection” tools the instructors possess be kept private. Once these tools are public knowledge, they become useless in detecting cheating. We have presented two specific cases with compelling cheating evidence during the Final Exam. Could it be true that cheating was not limited to these particular cases? The same version of the course was given several times in the past all with proctored in-class exams. Fig. 12 shows the students’ performance across exams since 2010. To compare the performance of students in Spring 2020 with past students who took the course, we use the following simple specification,(2) Scorei,s,j=α0+α1(Treat×Midterm2)s,j+α2(Treat×Midterm3)s,j+α3(Treat×Final)s,j+α4Treats+ηj+ϵi,s,j where Scorei,s,j is the exam score of student i in section s in exam j; Treats=1 if the section is from Spring 2020; Midterm2j, Midterm3j, Finalj are indicators for the corresponding exams; ηj is exam fixed-effects; ϵi,s,j is the idiosyncratic shock. Midterm 1 is taken as the baseline. It appears that more than two students may have cheated on the Final Exam. Table 6 shows how two sections from Spring 2020 performed across exams compared to the previous students who took the very same course. Column 1 compares 2020 Section 1 students to the past students. Similarly, column 2 compares 2020 Section 2 students to past students. Both sections performed worse on Midterms 2 and 3 compared to past students.17 However on the Final Exam, Section 1 outperformed the past students by approximately 4.3 points while Section 2 still performed slightly worse. This means that on average, a student in Section 1 received almost one higher letter grade on the Final Exam than their past counterparts.Table 6 Exam score differences comparing online vs. face-to-face delivery from past 10 years. Table 6outcome: exam score (1) (2) Section 1 Section 2 Midterm 2 -7.183*** -7.183*** (1.732) (1.732) Midterm 3 -3.382* -3.382* (1.617) (1.617) Final -8.033*** -8.033*** (2.116) (2.116) Treat -1.670 0.009 (2.019) (2.019) Treat×Midterm 2 -10.416*** -5.172** (1.732) (1.732) Treat×Midterm 3 -10.795*** -7.509*** (1.617) (1.617) Treat×Final 5.913** -1.640 (2.116) (2.116) N 1674 1682 Notes:Section 1 2020 students and Section 2 2020 students are compared with past 10-year students separately in columns (1)-(2). Baseline is Midterm 1, which was held face-to-face at the beginning of Spring 2020 before the COVID-19 related campus closures. Midterm 3 and the Final exam were held online. Students 1 and 2 were enrolled in Section 1. Treat is an indicator for the corresponding online section in 2020. Exam scores are out of 100 points. Clustered standard errors (clustered by section) are shown in parentheses. *p<0.1,**p<0.05,***p<0.01 These findings suggest that combined with incentives and peer-effects in cheating, there may also be a learning process in cheating. There appears to be no evidence of cheating on Midterm 3 - the first online exam - but evidence of cheating exists on the Final Exam, where gains are much more salient to students for any potential improvement in their grades. 4 Conclusion “I am talking about cheating. Unfortunately, no one can be trusted, except maybe for the top players for whom their reputation is the key asset.” - Arkady Dvorkovich, FIDE President, April 2020. (In response) “Wow. So sad. The biggest insult ever to all chess players just surfaced online.” - Jovan Retronic, International Chess Master, April 2020. Like doping, cheating cannot be completely eliminated. There always was, is, and will be cheating in face-to-face and online examinations. However, we can (try to) keep it at an expected equilibrium level. In this paper, we first looked at two simple models of face-to-face and online examinations. The theory suggests that cheating should be expected online. Then, we presented evidence of cheating that took place in an online examination in Spring 2020 under COVID-19 lockdowns and made suggestions on how to mitigate cheating based on the experience accumulated by online chess communities in the last two decades. COVID-19 made online chess much more popular since March 2020, and there is a growing number of online chess tournaments with substantial monetary prizes. This online chess experience is very similar to the experience of many academic instructors. The recent evidence suggests that the problem is not only there, but it is getting worse. In the intermediate Section B (1401–1700) of the recent European Online Chess Championship, 5 out of the top 6 players have been banned for cheating. The comment of International Grandmaster Nigel Short, FIDE Vice-President, on May 25, 2020 is revealing: “This scourge will not stop until people are criminally prosecuted for fraud.” Whether people indeed be prosecuted or not, one thing is clear: there is no chance to win a prize in an online chess event without proctoring if you do not cheat because you expect that everybody else will cheat and this belief is fulfilled in a bad equilibrium where everybody cheats. Of course, some people are disqualified, but not all. What does this mean for online exams? If instructors curve their grades, then they create a competition among students similar to what is seen in chess tournaments. Now, each student has more incentives to cheat because if they believe that the rest of the group is cheating, then they must demonstrate better than at least average class performance in order to pass the class. Thus, a student’s chance to pass the class without cheating would be very slim. Of course in this case, the cost of cheating goes down because the alternative to not cheating is failing the class. Therefore, any grade curving should not be used for online teaching. If universities decide to implement online exams with no proctoring in the upcoming semesters, we expect that there will be widespread cheating among students, who will not be penalized since it is almost impossible to present definitive evidence of cheating in an online exam. Unlike in online chess platforms, it will be difficult to implement our second suggestion for public universities that the instructors should be allowed to make the final decision for students who refuse to use a camera. Therefore, universities should implement a uniform online exam policy where a camera capturing each student’s computer screen and room is a requirement. A camera will also help to check a student’s ID and eliminate the possibility of another person taking the test.18 For instructors, in addition to not curving any grades, we also suggest to give students less time but easier questions to increase the value of time, making it more costly to cheat. Fig. 8 shows a rise in cheating in the world’s largest online chess platform since shutdowns due to COVID-19 began. Today, several online chess tournaments have a policy that requires participants to use a camera to live-stream and record themselves during the tournament.19 Fig. 9 shows such setup used in a recent online chess tournament.20 Players live-stream from a side-angle camera showing their screen and surroundings with their microphone enabled. Even though this method cannot eliminate all cheating, we believe it is a great balance between having no proctoring and using online proctoring services.21 Fig. 8 Account closures on Chess.com since shutdowns due to COVID-19 began Note: Source: https://www.chess.com/article/view/online-chess-cheating#false-positives. Fig. 8 Fig. 9 Playing chess on Chessking.com. Note: Source: https://ruchess.ru/news/report/sokhranyaya_distantsiyu. Fig. 9 Of course, the problem is much bigger, as was noted by Peter Heine Nielsen, Coach of World Chess Champion Magnus Carlsen, on May 25, 2020: “The same could be said about corruption, pre-arranged games, buying of votes, jobs going to friends or political allies instead of an open recruitment procedure based on merits etc. These are big issues for the chess world, not a 1400–1700 online event.” Declaration of Competing Interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Appendix A Fig. 10 Comparison of gains in scores for students between exams. Note: Midterms 1 and 2 were held face-to-face with proctoring; Midterm 3 and the Final Exam were held online asynchronously without proctoring following COVID-19 related campus closures. Each bar represents a student’s test score gains between corresponding exams. Each exam is worth 20 points. Fig. 10 Fig. 11 Comparison of gains in scores for students across exams. Note: Midterms 1 and 2 were held face-to-face with proctoring; Midterm 3 and the Final Exam were held online asynchronously without proctoring following COVID-19 related campus closures. Each dot represents a student’s performance in corresponding exams. The dashed line is the 45 degree line. Fig. 11 Fig. 12 Performance on exams across years (for students who took the final exam) Notes: All exams before 2020 were held face-to-face with in-class proctoring. Midterms 1 and 2 in 2020 were held face-to-face at the beginning of Spring 2020 before the COVID-19 related campus closures. Midterm 3 and the Final exams in 2020 were held online. Fig. 12 ☆ We would like to thank Jan Bass, Philip Brookins, Bentley Coffey, Joshua Hess, Chun-Hui Miao, Edsel Pena, Tamara Sheldon, Lindsey Woodworth, and seminar participants at the University of South Carolina for their questions and suggestions. Special thanks to the Office of Student Conduct and Academic Integrity at University of South Carolina, especially to Erin Kitchell, Director of Academic Integrity, for her feedback and comments. ⁎⁎ The International Chess Federation (FIDE) is the governing body of chess, and it regulates all international chess competitions. 1 See the Washington Post article and the article on InsideHigherEd.. 2 College Board did not consider internet search to be cheating for the 2020 AP examinations. However, even if internet search was considered cheating, ensuring that students not use internet search during the test would be a challenging task. 3 Cheating in chess is a big issue in both online and over-the-board settings. This problem is relevant even in scholastic chess events. See https://en.chessbase.com/post/promoting-fair-play-among-child-chess-players. 4 https://www.chess.com/article/view/chess-com-fair-play-and-cheat-detection. 5 See https://www.chess.com/article/view/online-chess-cheating. 6 See https://www.spraggettonchess.com/the-game-of-cheating-part-i/. 7 For example, there are numerous cases of titled players admitting they had cheated and were correctly identified and caught by Chess.com. 8 Along with Jacob and Levitt (2003) these papers were later included in Levitt and Dubner (2005): the Freakonomics book, documentary, and podcast series. 9 In fact, the game can be played sequentially without the professor knowing the student’s action. The normal-form of this game and our results are the same. 10 See our discussion in Section 2. 11 Anectodal evidence suggests that many instructors were indeed reluctant to report cheating in Spring 2020. Despite this, the number of reported cheating cases at a large public university (reported by their Academic Integrity Office) went up by almost 10% in March - June 2020 relative to March - June 2019. 12 The questions on the exam are problem-solving questions which are arguably not-so-trivial in terms of finding the solutions. The exam is not multiple-choice – the student must type in the correct answer to receive credit. 13 There were instances of students finishing their tests and waiting to submit them. The test was designed such that students could not go back and recheck their answers. Therefore, waiting could not improve or change their results. In one extreme case, a student finished the test in 11 min and waited for more than 1 h before submitting it, so that the total time spent on the test would look “normal”. We believe this provides evidence on individuals involved in cheating attempting to “hide their trails” similar to what was observed in Jacob and Levitt (2003). 14 A rough estimate on the probability of “being lucky” on the Final Exam and guessing the correct answers by randomly submitting numbers is less than 1×10−20. The probability of winning the lottery is around 1×10−7. This rough estimate takes into account a student’s potential to “guesstimate” the range for the correct answer. 15 This question was by far the most difficult question on the exam with a correct response rate of only 19.3%. The third student who submitted “125” scored 18/20 on the Final Exam. His other incorrect answer was “12”, and the correct answer for that question was “124”. It appears he simply must have made a typo. 16 It would potentially take 15–20 seconds to identify what question comes up on the screen and match it with the solution key they have at hand since the order of the questions is randomized for each student. 17 The exams were designed such that the difficulty of each exam goes up moving from the first Midterm to the Final Exam. In addition, all past students received multiple choice questions while the students from Spring 2020 received a similar set of questions with no multiple choice options, but instead were asked to type in their answers. 18 See a discussion on this issue https://www.michigandaily.com/section/academics/university-faculty-and-students-discuss-academic-integrity-digital-classroom. 19 See the regulations for a recent online chess event held in Spring 2020 and the regulations for an upcoming online chess event to be held in Fall 2020. 20 Must be on Zoom (use real name) to be eligible for prizes (side/rear camera angle). 21 Online proctoring services are often associated with privacy concerns related to the use and storage of personal data. These services are also criticized for their use of the “AI” and created petitions against using them. See among many articles https://www.cnn.com/2020/08/29/tech/online-school-test-surveillance/index.html, https://www.dailymail.co.uk/news/article-8243637/Creepy-software-used-stop-university-students-cheating-online-exams-amid-coronavirus.html and https://www.vox.com/recode/2020/5/4/21241062/schools-cheating-proctorio-artificial-intelligence. ==== Refs References Alan S. Ertac S. Gumren M. Cheating and incentives in a performance context: evidence from a field experiment on children J. Econ. Behav. Organ. 179 2020 681 701 10.1016/j.jebo.2019.03.015 Becker G.S. Crime and punishment: an economic approach J. Polit. Econ. 76 2 1968 169 217 Charness G. Blanco-Jimenez C. Ezquerra L. Rodriguez-Lara I. Cheating, incentives, and money manipulation Exp. Econ. 22 1 2019 155 177 10.1007/s10683-018-9584-1 Dee T.S. Jacob B.A. Rational ignorance in education: a field experiment in student plagiarism J. Hum. Resour. 47 2 2012 397 434 Diedenhofen B. Musch J. Pagefocus: using paradata to detect and prevent cheating on online achievement tests Behav. Res. Methods 49 4 2017 1444 1459 10.3758/s13428-016-0800-7 27573006 Duggan M. Levitt S.D. Winning isn’t everything: corruption in sumo wrestling Am. Econ. Rev. 92 5 2002 1594 1605 10.1257/000282802762024665 Erat S. Gneezy U. White lies Manage Sci 58 4 2012 723 733 10.1287/mnsc.1110.1449 Fask A. Englander F. Wang Z. Do online exams facilitate cheating? An experiment designed to separate possible cheating from the effect of the online test taking environment J. Acad. Ethics 12 2 2014 101 112 10.1007/s10805-014-9207-1 Fischbacher U. Föllmi-Heusi F. Lies in disguise-an experimental study on cheating J. Eur. Econ. Assoc. 11 3 2013 525 547 10.1111/jeea.12014 Gächter S. Schulz J.F. Intrinsic honesty and the prevalence of rule violations across societies Nature 531 7595 2016 496 499 10.1038/nature17160 26958830 Gneezy U. Deception: the role of consequences Am. Econ. Rev. 95 1 2005 384 394 10.1257/0002828053828662 Jacob B.A. Levitt S.D. Rotten apples: an investigation of the prevalence and predictors of teacher cheating Q. J. Econ. 118 3 2003 843 877 10.1162/00335530360698441 Karim M.N. Kaminsky S.E. Behrend T.S. Cheating, reactions, and performance in remotely proctored testing: an exploratory experimental study J. Bus. Psychol. 29 4 2014 555 572 10.1007/s10869-014-9343-z Lederman D. Best way to stop cheating in online courses? each better Inside Higher Ed 2020 https://www.insidehighered.com/digital-learning/article/2020/07/22/technology-best-way-stop-online-cheating-no-experts-say-better Levitt S.D. Dubner S.J. Freakonomics: A Rogue Economist Explores the Hidden Side of Everything 2005 William Morrow & Co New York, NY, US Maggioni M.A. Rossignoli D. Clever little lies: math performance and cheating in primary schools in congo J. Econ. Behav. Organ. 172 2020 380 400 10.1016/j.jebo.2019.12.021 Martinelli C. Parker S.W. Pérez-Gea A.C. Rodrigo R. Cheating and incentives: learning from a policy experiment Am. Econ. J.: Econ. Policy 10 1 2018 298 325 10.1257/pol.20150066 Newton, D., 2020. Another problem with shifting education online: a rise in cheating. Washington Post. https://www.washingtonpost.com/local/education/another-problem-with-shifting-education-online-a-rise-in-cheating/2020/08/07/1284c9f6-d762-11ea-aff6-220dd3a14741_story.html. Vanberg C. Who never tells a lie? Exp. Econ. 20 2 2017 448 459 10.1007/s10683-016-9491-2 Watson G. Sottile J. Cheating in the digital age: do students cheat more in online courses? Online J. Distance Learn. Admin. 13 1 2010

0

PMC9754215

NO-CC CODE

2022-12-16 23:26:31

no

J Econ Behav Organ. 2021 Feb 28; 182:196-211

utf-8

J Econ Behav Organ

10.1016/j.jebo.2020.12.004

oa_other

==== Front Gen Hosp Psychiatry Gen Hosp Psychiatry General Hospital Psychiatry 0163-8343 1873-7714 Elsevier Inc. S0163-8343(21)00076-1 10.1016/j.genhosppsych.2021.05.006 Letter to the Editor A national snapshot of U.S. adolescents' mental health and changing technology use during COVID-19 Burke Taylor A. a Kutok Emily R. b Dunsiger Shira c Nugent Nicole R. a Patena John V. b Riese Alison d Ranney Megan L. b⁎ a Department of Psychiatry and Human Behavior, Brown University, 700 Butler Drive, Providence, RI 02906, United States b Brown-Lifespan Center for Digital Health, 139 Point Street, Providence, RI 02903, United States c Department of Behavioral and Social Sciences, Brown University, Box G-5121-4, Providence, RI 02912, United States d Department of Pediatrics, Alpert Medical School of Brown University, 593 Eddy Street, Potter 200.9, Providence, RI 02903, United States ⁎ Corresponding author at: Brown-Lifespan Center for Digital Health, 139 Point St, Providence, RI 02903, United States. 25 5 2021 July-August 2021 25 5 2021 71 147148 6 5 2021 21 5 2021 24 5 2021 © 2021 Elsevier Inc. All rights reserved. 2021 Elsevier Inc. Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. Keywords COVID-19 Social media Technology Mental health Adolescents ==== Body pmcPreliminary reports suggest that during COVID-19, adolescents' mental health has worsened while technology and social media use has increased. Much data derives from early in the pandemic, when schools were uniformly remote and personal/family stressors related to the pandemic were limited. This cross-sectional study, conducted during Fall 2020, examines the correlation between mental wellbeing and COVID-19-related changes in technology use, along with influence of COVID-19-related stressors, school status (in-person versus remote), and social media use for coping purposes, among U.S. adolescents. From September 23 to December 16, 2020, English-speaking adolescents (ages 13–17) residing in the United States were recruited using Instagram for an online survey, with approval from the Institutional Review Board. Assent was waived, with approval from the Institutional Review Board. Self-report measures (adapted from Pew Internet Survey [1]) assessed average daily duration of technology use (social media, phone/video calls, video games, TV/movie/videos) 30 days before initial COVID-19-related school closures versus past week. Standard measures for past week anxiety and depressive symptoms (PROMIS) [2], well-being (WHO-5) [3], and cybervictimization [4] were used. Use of social media for coping through social connection was assessed using an adapted measure for the purpose of the present study. School status (open full-time or hybrid versus closed) was determined through the use of the COVID-19 US State Policy Database [5]. COVID-19-related stressors [6], perceived importance of social media [7], and demographics were also assessed. Generalized linear models were used to examine associations between changes in technology use and current mental health outcomes, adjusting for COVID-19-related stressors and importance of social media (identified as confounders in preliminary analysis); potential moderators were examined. We recruited 978 youth from 42 states (Supplementary Table 1). All forms of technology use significantly increased from pre-COVID until the time of assessment (Supplementary Table 2). After adjustment for confounders, self-reported increases in social media use were associated with higher anxiety and depressive symptoms (Table 1 ). The extent of use of social media for coping through social connection moderated the association between social media use and depressive symptoms (b = 0.15, SE = 0.07, p = .02). Results indicated that among those who report infrequent use of social media for coping, greater increases in social media use were associated with higher depressive symptoms (b = 0.16, SE = 0.07, p = .02). However, among those who report frequent use of social media for coping, there were no associations between changes in use and depressive symptoms (b = 0.04, SE = 0.06, p = .48). Increases in video gaming and TV/movie watching were also associated with higher depressive symptoms, and video gaming was associated with higher anxiety.Table 1 Adjusted effects of changes in technology use on mental health outcomes. Table 1 Anxiety symptoms Depressive symptoms Well-being Cybervictimization Change in time on social media b = 0.07, SE = 0.03⁎ b = 0.11, SE = 0.03⁎ b = −0.21, SE = 0.13 b = −0.06, SE = 0.08 Change in time on phone or video calls b = 0.02, SE = 0.03 b = 0.02, SE = 0.03 b = 0.07, SE = 0.12 b = −0.08, SE = 0.07 Change in time on video games b = 0.09, SE = 0.03⁎ b = 0.06, SE = 0.03⁎ b = −0.03, SE = 0.14 b = 0.12, SE = 0.09 Change in time on TV, movies, videos b = 0.06, SE = 0.03 b = 0.10, SE = 0.03⁎ b = −0.05, SE = 0.15 b = −0.06, SE = 0.09 Note. Changes in technology time reflect differences: past 7 days – one month before school closures; Models adjusted for COVID-19-specific stressors and importance of social media; b = unstandardized regression coefficient; SE = standard error. ⁎ p < .05. There were no associations between changes in any form of technology use and overall well-being or cybervictimization. Neither local school status, nor level of COVID-19-related stressors, nor self-perceived importance of technology, were significant confounders or moderators of the observed effect. In this geographically diverse sample of adolescents across the United States, self-reported daily social media and technology use increased significantly from prior to COVID-19 through Fall 2020. Increased social media use was significantly associated with higher levels of anxiety and depressive symptoms regardless of other theoretical moderators or confounders of mental health (e.g., demographics, school status, importance of technology, COVID-19-related stress). Despite literature suggesting that remote learning may result in adverse mental health outcomes [8], we did not find local school reopening to be associated with current depressive/anxiety symptoms, nor with COVID-19-related increases in technology use. Self-reported use of social media for coping purposes moderated the association between increased social media use and depressive symptoms, such that an association between these constructs was found only for individuals who infrequently use social media for coping purposes. It is therefore possible that greater use of social media for certain purposes may have protective effects [9]. Although much prior research has focused on social media use as a marker of stress, we also found that increased video gaming and TV/movie watching were also associated with internalizing symptoms, in accordance with others' work [10]. Future research should explore in more granular detail what, if any, social media and technology use is protective during a pandemic, and for whom, to help tailor prevention efforts. Importantly, the use of a cross-sectional design limits our ability to disentangle the directionality of associations between technology use and mental health symptoms. Additional limitations include use of some non-validated measures, reliance on self-report of technology use, and use of a national database to assess school status. In conclusion, our study shows that, although adolescents' self-reported technology use increased from prior to the pandemic until Fall 2020 and was associated with poorer mental health, the relationship may be more nuanced than previously reported. Funding/support This study was funded by a grant from the Technology and Adolescent Mental Wellness (TAM) program at the 10.13039/100007015 University of Wisconsin-Madison , grant 0000000136/132580194. Role of the funding source The Technology and Adolescent Mental Wellness (TAM) program had no role in the design and conduct of the study. Author contributions Dr. Ranney and Dr. Burke conceptualized and designed the study, drafted the initial manuscript, and reviewed and revised the manuscript. Dr. Dunsiger carried out the initial analyses and drafted the results section, and reviewed and revised the manuscript. Ms. Kutok collected data and drafted the methods section, and reviewed and revised the manuscript. Dr. Riese, Dr. Nugent, and Mr. Patena reviewed and revised the manuscript. All authors approved the final manuscript as submitted and agree to be accountable for all aspects of the work. Conflicts of interest disclosures Dr. Ranney holds stock in Moderna and has received money for consultation from Medscape for talks on COVID-19 testing. Drs. Ranney, Dunsiger, and Nugent have NIH and CDC grants for other projects. Appendix A Supplementary data Supplementary material Image 1 Acknowledgments The content is solely the responsibility of the authors and does not necessarily represent the official views of the university or the TAM program. Taylor Burke was supported by 10.13039/100000025 NIMH T32 MH019927. Appendix A Supplementary data to this article can be found online at https://doi.org/10.1016/j.genhosppsych.2021.05.006. ==== Refs References 1 Lenhart A, Smith A, Page D. Teens, technology and romantic relationships. https://www.pewresearch.org/internet/2015/10/01/teens-technology-and-romantic-relationships/ Accessed February 23, 2021, 2021. 2 Irwin D.E. Stucky B. Langer M.M. An item response analysis of the pediatric PROMIS anxiety and depressive symptoms scales Qual Life Res 19 4 2010 595 607 10.1007/s11136-010-9619-3 20213516 3 Regional Office for Europe WHO Use of well-being measures in primary health care. The DepCare project health for all, Target 12 Published online 1998 1998 45 https://www.euro.who.int/__data/assets/pdf_file/0016/130750/E60246.pdf 4 Jones L.M. Mitchell K.J. Defining and measuring youth digital citizenship New Media Soc 18 9 2016 2063 2079 10.1177/1461444815577797 5 Raifman J. Nocka K. Jones D. Bor J. Lipson S. Jay J. COVID-19 US state policy database 2020 6 Nikolaidis A. Paksarian D. Alexander L. The Coronavirus Health and Impact Survey (CRISIS) reveals reproducible correlates of pandemic-related mood states across the Atlantic Sci Rep 11 2021 8139 10.1038/s41598-021-87270-3 7 Rideout V. Robb M.B. Social life media: 2018 teens reveal their experiences 2018 Common Sense Media 10.1016/j.ijheatmasstransfer.2016.02.015 8 Lee J. Mental health effects of school closures during COVID-19 Lancet Child Adolescent Health 4 6 2020 421 32302537 9 Allen K.A. Ryan T. Gray D.L. McInerney D.M. Waters L. Social media use and social connectedness in adolescents: the positives and the potential pitfalls Aust Educ Dev Psychol 31 1 2014 18 31 10 Lobel A. Engels R.C.M.E. Stone L.L. Burk W.J. Granic I. Video gaming and children’s psychosocial wellbeing: a longitudinal study J Youth Adolesc 46 4 2017 884 897 28224404

34090695

PMC9754300

NO-CC CODE

2022-12-16 23:26:35

no

Gen Hosp Psychiatry. 2021 May 25 July-August; 71:147-148

utf-8

Gen Hosp Psychiatry

10.1016/j.genhosppsych.2021.05.006

oa_other

==== Front Sex Cult Sex Cult Sexuality & Culture 1095-5143 1936-4822 Springer US New York 10049 10.1007/s12119-022-10049-9 Original Article “Coming Out To Yourself”: Reflections On Early-Years Sexual Identity Formation Among Different Generations of Bulgarian Non-Heterosexual Males http://orcid.org/0000-0002-4611-6186 Darakchi Shaban shaban.darakchiev@gmail.com 12 1 grid.5284.b 0000 0001 0790 3681 University of Antwerp, Antwerp, Belgium 2 grid.410344.6 0000 0001 2097 3094 Bulgarian Academy of Sciences, Sofia, Bulgaria 15 12 2022 121 23 7 2022 26 11 2022 © The Author(s), under exclusive licence to Springer Science+Business Media, LLC, part of Springer Nature 2022, Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. During the past few years the so-called “anti-gender campaigns” in Bulgaria have revitalized the polemics surrounding the development of non-heterosexual identities claiming that these identities are “imported” by “Western” politics and discourses in order to “weaken” and transform national cultural and political models. Analyzing 63 semi-structured in-depth interviews with non-heterosexual males from different generations, this study aims to contribute to the theories of non-heterosexual identity development by providing data from Bulgarian context. The data from this study suggests that: (1) non-heterosexual male identities in Bulgaria have existed before the “global gay culture”; (2) the younger the participants the earlier they realize their non-heterosexual desires often within the “pre-sexuality stage” defined by the stage models and the youngest cohort self-label their same-sex attraction mainly through an “identity-centred” sequence, before engaging in sexual activities; (3) the greater awareness of role models, the wider access to information, and the involvement in the LGBTQI + communities have contributed to a more positive and self-respectful identity development; (4) physical contacts and observations as significant sources for the questioning of a non-heterosexual identity have been replaced by virtual observations and communication; (5) the Internet and social media have made non-heterosexual identity development more accessible regardless of social and economic background, and that (6) non-heterosexual identity development does not lead automatically to a culturally defined gay identity. Keywords Sexuality Non-Heterosexual Gay Identity Anti-Gender Campaigns Homonormativity Eastern Europe Research Foundation Flanders FWO12Z7320N Darakchi Shaban ==== Body pmcIntroduction The debates on the “nature” of homosexuality have been significantly revitalized during the backlash known as “anti-gender campaigns” in recent years. The proponents of the anti-gender movements claim that homosexuality “can be learned and taught” and the so-called “gender ideology” is an “instrument” for teaching homosexuality to youngsters (Korolczuk & Graff, 2018) Therefore, all educational activities that mention anything remotely connected to gender or sexuality have been one of the main targets of the anti-gender movements. These processes have taken place in many Eastern European countries, Brazil, Colombia and others (Kuhar & Paternotte, 2017). These campaigns have been extremely successful in Bulgaria. In a series of campaigns on social media maintained by certain religious and nationalistic organizations and parties, any educational policies which mention “gender” or LGBT have been attacked and the people involved in these policies have been publicly humiliated and threatened. Furthermore, in these discourses homosexuality is viewed as something “imported, not existing before 1989 and portrayed as an instrument for invasion of poorer countries by the “West” by making them “less populated” (Darakchi, 2019). The qualitative studies investigating the “lived” and often hidden experiences of non-heterosexual individuals remain a small portion of the studies devoted to sexuality and sexual identity. These recent events and the polemics surrounding the “nature” of homosexuality as “imported” in Bulgaria require a detailed investigation of sexual identity formation in the former communist countries. To date, there is no single qualitative study investigating the formation of non-heterosexual identities in Bulgaria. Thus the local stories and lived experiences are missing in these discussions; instead, the development of non-heterosexual identity is told predominantly by the populist far-right movements and public figures. This study aims to provide insight into the formation of non-heterosexual male identities in Bulgaria by using an intergenerational perspective and semi-structured in-depth interviews. Responding to the need for qualitative data investigating the awareness of one’s sexual orientation, this paper focuses on the early years experiences of the participants and seeks to answer three main questions. First, how has non-heterosexual orientation awareness happened in the Bulgarian context during different historic periods? Second, which are the main self-identified milestones in the development of a non-heterosexual identity? Third, which are the sources of information that have influenced the formation of non-heterosexual identity over time? Theoretical Approaches The first models to investigate the development of sexual identity were the so-called “stage models”. The stage models define specific, “universal” stages which constitute the development of sexual identity (Bilodeau & Renn, 2005). Many studies have found discrepancies between individual experiences and the “stage models” (Olive, 2012). Furthermore, the “stage models” suggest that the individuals will finally “come out” to their environment. However, the Internet nowadays allows for anonymous coming out and coming out to geographically distant people or networks (Giano, 2019) and besides some individuals might never “come out”. Criticizing “the stage models” as limited, D’Augelli (1994) proposed the model of sexual development as a “life span” process that considers setbacks and nonconsecutive stages within the individual experiences. Similarly, Lipkin (2001:103) proposed a “mega-model” that combines the previous most cited stage models and argues for flexibility when it comes to the consecutive stages. Lipkin’s model consists of 5 stages: Pre-Sexuality: Preadolescent nonsexual feelings of difference and marginality; Identity Questioning: Ambiguous, repressed, sexualized same-gender feelings and/or activities. Avoidance of stigmatized labels; Coming Out: Toleration then acceptance of identity through contact with gay/lesbian individuals and culture. Exploration of sexual possibilities and first erotic relationships. Careful, selective self-disclosure outside the gay/lesbian community. Pride: Integration of sexuality into self. Capacity for love relationships. Wider self-disclosure and better stigma management. Post-Sexuality: A diminishment of the centrality of homosexuality in self-concept and social relations. Lipkin (2001) noted that this mega-model is a “gross generalization” and does not represent all the experiences; however, it can be used as a framework when investigating the development of non-heterosexual identity. This article focuses on the first 3 “stages” of the model, emphasizing the milestones which influence one to come out to oneself rather than to come out to “the others”. The “stage models” have emphasized the structures which contribute to one’s self-awareness of sexual desires and sexual identity. The structures in a certain society or a certain historical period however can only present a limited picture of different realities and different experiences and fail to provide sufficient information on collective experiences typical for a specific cohort (Cohler & Hammack, 2006). One possible solution to overcome this limitation is to “reclaim the gay past” (Duberman, 1988), using the individuals’ voices (Dowsett, 1996; Seidman 2004). This allows for a better reconstruction of the socio-economic conditions which have played roles in the development of a non-heterosexual identity among the studied group and, in particular, historic periods and specific political and economic conditions (Parker, 1989). At the same time, it also allows registering the differences and changes in the reconsideration, self-management and labelling of the “sexual” (Coleman-Fountain, 2014). Overcoming the dualism between structural and constructivist approaches, combining the “objective” and the “subjective”, the material and the cultural (Husu, 2013) is specifically useful in the Bulgarian context where the knowledge and the archives from the communist past before 1989 remain very limited. Moreover, the communist period is usually associated with the “lack” of same-sex practices and relationships in many public narratives. Contrary to this “belief,” Chauncey (2008) demonstrated that behind the myths of the nonexistence of gay cultures and practices there was a proliferating gay world hidden from the public. Early Years’ non-heterosexual Identity Formation Different sociohistorical circumstances define different possibilities for people with same-sex attraction and life stories hold the power to offer different “representations of identity” (Cohler & Hammack, 2006). A very important period in this process is the years before adolescence (up to 10) and the years of adolescence when the young individual discovers and reflects on their desires and allocates them to the available categories in a specific cultural context (Driver, 2008). In many societies across the globe, the existing categories are usually heteronormative and the structures do not allow any forms of non-heterosexual being (Coleman-Fountain, 2014). In a heteronormative society, children interact and grow up within the so-called “heterosexual market” (McConnell-Ginet & Eckert, 2003). This is a social organization of language, rituals, festivities, clothes, colors and other categories which allocate specific roles to boys and girls preparing them to become couples. This “market” is especially visible in all the school activities. As a result of this, some adopt a strategy of outperforming their peers academically and in sports to compensate for their inability to fit into the “market” (Lipkin, 2001; Fuentes, 2020). The resources available to young people and society’s role models are crucially important in this period (Dube, 2000; Driver, 2008). Sexual Encounters, Gender Performativity and non-heterosexual Identity Formation The first sexual encounter is a major milestone in the development of non-heterosexual identity. Some recent studies reported that the lack of proper educational and cultural settings leads to negative experiences, unpreparedness and a lack of proper language among LGBTQI + individuals during their first sexual encounters (Gillespie et al., 2021). Furthermore, in a heteronormative non-heterosexual environment any sexual practice or preference which does not conform to the idea of “normality” may be considered ‘bad” and may be subject to criticism, “normalization” or rejection. Usually the notion of the “good” citizen frames the sexual with the romantic, the relationship, the marriage and children, and the monogamy – all typical for the heteronormative structures (Seidman, 2004). The first sexual encounter is for many non-heterosexual males a milestone when they reconsider and rediscover their physical and emotional preferences (Dube, 2000). This involves one’s understanding of masculinity and performance (Connell, 1991; Shio & Moyer, 2021), where common strategies are dating females (Klinkenberg & Rose, 1994) and strait-acting performances to different audiences (Eguchi, 2009; Edwards, 2012). The first sexual encounters among non-heterosexual males are experienced more traumatically than those among non-heterosexual females due to the patriarchal norms (Hegna & Larsen, 2007). This is especially valid in patriarchal societies where the receptive role is viewed as feminine and the penetrative role as masculine (Parker, 1989; Bereket & Adam, 2006). Furthermore, gay pornography has been portraying features of male dominance and a certain image of the “male” which is often incorporated as a “standard” among non-heterosexual males. These stereotypes however are not stable. The heteronormative structures and scripts have been tackled by different audiovisual agencies and social activism bringing up on camera diverse male bodies and behaviors (Rothmann, 2013). Moreover, some individuals identify as non-heterosexual before any sexual encounters (Dube, 2000) as a result of greater social awareness and access to information. This study focuses not only on the first same-sex encounters but also seeks to understand the self-reflection of those who compare their first sexual encounters with females to those with males as a process of sexual identity development. “Discovering” non-heterosexual Identity Globalization and technologies have led to significant shifts in the social construction of non-heterosexual identities. Education and awareness of alternative lifestyles have contributed to greater freedom and individual choices, which have resulted in major conflicts between local normative structures and mythologies on one side and new technologies and ideologies on the other (Altman, 1996; Parker, 1989). The rapid transformation and globalization in the past few decades have brought a series of reconsiderations which require an investigation of sexual identity development among non-heterosexual people as a life-long process (Floyd & Stein, 2002). Sexual identity is “discovered” through discourse and conversations, reading and self-reflection on one’s sexual desires and experiences. The development of non-heterosexual identity in this regard is an interaction and interplay between different sources of information such as books, movies, newspapers, and internet sources. The interpretations and reflections on this information (Cohler & Hammack, 2006) constitute sexual objectification, self-objectification and subjectification (Dowsett, 2015). According to Martel (2018) “satellite TV, mobile screens, internet, and social networks” have immensely transformed the lives of LGBT people across the globe. The internet has created many possibilities for the development of non-heterosexual identity and many studies suggest that online interactions, storytelling and acquaintances have predefined the development of non-heterosexual identity providing anonymity and wider access to information and support (Giano, 2019). Gay dating websites and applications have further widened the possibility to connect and communicate anonymously without being a part of a community and without coming out “officially” (Mowlabocus, 2016). Many studies have focused on exploring the structures and messages of the available content, such as movies (Seidman, 2004), music, arts, internet pages and other sources, to discover what might have influenced the development of non-heterosexual identities. What I am interested in is the availability of these resources in Bulgaria as well as the participant’s reflections, objectification and subjectification of this information and sources. Method This study uses a qualitative research methodology. In-depth interviews (semi-structured questionnaire) combined with a narrative approach assure the “trajectory of life across time”, depth and coherence of the accounts (Carless & Douglas, 2017). I interviewed 63 self-identified non-heterosexual males in the period June 2020 – April 2021 following sampling procedures in previous studies (Harry, 1986: Merriam, 2002). All the participants chose the place for the interview: 43 people chose different public settings (restaurants; bars, parks); 13 people were interviewed online due to Covid-19 measures, and 8 interviews took place in my home. The interviews lasted from 1 h to 34 min to 4 h and 47 min. The participants’ names were anonymized. I used a combined sampling procedure. For the initial contact with different respondents, I consulted the LGBTQI + organizations GLAS, Bilitis and Deystvie and my networks. This is how I got into contact with 8 people of diverse backgrounds and community involvement. A snowballing procedure based on the initial contacts put me in contact with additional 13 people. Based on the recommendations given by the last group I made contacts with additional 22 people. I contacted the remaining 20 participants directly on Facebook after some observation of the comment sections on two Bulgarian LGBTQI + Facebook groups taking into account diverse opinions and demographic statuses. The respondents represent diverse groups in terms of age, place of living, ethnicity and education. Regarding age: 18–25 years old – 8 participants; 25–30 years old − 9; 30–40 years old − 18; 40–50 years old − 17; 50–65–7; above 65 years old – 4. Identifying generational similarities based on interactions between historical events and personal experiences in studies devoted to non-heterosexual people is a challenging task given the variety of subjective experiences (Dhoest, 2022). The generations identified in other studies (Cohler & Hammack, 2006; Dhoest, 2022) based primarily on key historic events such as liberation movements, Stonewall riots, HIV/AIDS crisis and others, would be rather irrelevant in the Bulgarian context due to the isolation of Bulgaria from the “Western world” before 1989 during communism. For the purposes of the analysis, based on the data and taking into account the small number of people born before 1975, this study will distinguish between: Generation 1 (G1) - those born before 1980 who came to terms with their sexuality during communism with very limited access to information and impossibility for open self-expression. Age remains the biggest challenge to the sampling of participants since people above 65 years old would not readily agree to participate. It poses a certain risk for overestimations and misrepresentations in the analysis. Generation 2 (G2) - those born between 1980 and 1995 who had access to books, magazines, pornography, TV programs and later the Internet during their coming out of age. On the one hand, this generation grew up during the first decades of “democracy”, the emergence of the first gay and lesbian organizations, including the first gay pride, and the accession of Bulgaria to the EU. On the other hand, this was a period of “legitimization” of the nationalistic parties and the religious institutions which contested the liberation of sexual freedom. Generation 3 (G3) – the participants born after 1995 who grew up in times of expanding Internet and unlimited access to social media, movies, international mobility and increasing involvement in LGBTQI + activism and networks. Regarding the place of living, I selected an almost equal number of people from each age cohort living in small towns, regional cities and the capital – Sofia. Regarding ethnic identity, 51 people self-identified as ethnic Bulgarians, 5 as ethnic Turks, 2 as ethnic Armenians, 2 as ethnic Jews and 5 as ethnic Roma. Sexual orientation identifications vary as follows: 28 have used the term “gay”, 31 have used terms such as homosexual, MSM, and “simply male” (most of them distancing themselves from the term “gay”), and 4 respondents have identified as queer. When it comes to education status, it must be noted that a higher educational status does not automatically guarantee a better-paid or more highly qualified job. Some 49 people reported a higher educational degree (at least a bachelor’s – 4 years degree), 8 high education and 4 with primary education. Of those 49 people who reported a higher education degree, 27 were the first generation to obtain a higher degree diploma. The figures confirmed that higher education does not always result in a well-paid job position. Some of the respondents with higher degrees were manual workers such as waiters, cleaners or cooks. Due to the wide access to comparatively cheap higher education, a very high proportion of the people in Bulgaria have obtained a higher degree. On the one hand, the labor market cannot provide enough opportunities for all, on the other hand, people in some professions such as waiters earn more than those working in public administration, state schools and even hospitals in certain cases. The data were analyzed with NVivo research software. I transcribed all the interviews as a precaution that this information would not end up in inappropriate hands and threaten the participants’ well-being. Using thematic analyses, I outlined the main patterns and milestones discussed by the respondents. I used discourse analysis to identify the relations between the patterns and investigate the dynamics in these patterns over time. The main limitation of this study is related to accessibility and representation. Although I searched for participants from diverse backgrounds, I might not have included experiences from hard-to-reach groups such as people who refused to be interviewed (6 people) and they might represent models and patterns which were not included in the following analysis. Another major limitation is the small number of people born before 1975 willing to give an interview. This might also have led to a limited “restoration” of the past. Results “Becoming” non-heterosexual: Feeling “the difference” Very often in public discussions in Bulgaria, the non-heterosexual identity is considered as a sudden “becoming” due to media or other types of external influence. The data suggested that most of the participants felt “different” from a very early age according to their descriptions. This feeling of “being different” was experienced in different ways and was usually reported as first experienced between 7 and 12 years old. Nearly two-thirds of the participants shared that they used to be very shy children who did not have many friends and often played alone. “I had no idea what was wrong with me but I knew there was something, you feel it, it has to be felt, it is like in a horror movie, you expect that something bad is approaching” - Asen (31). The “difference” was often connected to playing with “female” toys, wearing female clothes and imitating female singers. “I have a clear memory with my dad when I wanted dolls but he insisted on buying me dinosaurs and bought me a dinosaur. When we got home I put some lipstick on the dinosaur, and some dresses” - Dani (37). Another major “difference” was felt by body comparisons. I must have been very feminine because I was always ridiculed for my gestures, I was also a bit fat, and they also named me after a female name. I did not like the male games, it was always aggressive for me, I remember I could not throw like them, my voice, I can say now, was not so masculine, I could not lift heavy stuff like them, I did not belong there”. - Alex (36). A significant milestone in self-understanding as “different” took place when the participants entered primary school. The school was usually described as aggressive and hostile towards the differences. Some 41 people reported abuse in their school years based on their looks and/or behavior. Mihail is 24 years old, born in a small town and he still “has chills” when he thinks about the time in school. He was constantly abused verbally and physically, called different female names and threatened. He preferred to stay home and read books most of the time. He never reported that to teachers or parents because his father was also controlling of his behavior and mannerisms. Mihail’s situation was common for many people in this study. Another very common feature of school life was the experience with sports. Once they were bullied in a school environment, the “difference” felt by the participants was taken to the football pitch. While there were other games such as volleyball where the participants felt uncomfortable, the football game appeared to be one of the most hostile situations in schools. Some 51 participants reflected on the football game as a very competitive, rude and unfriendly environment. These sentiments were also reported by many who would not consider themselves physically different. The football game appeared to be a situation where the physical abilities to play properly were judged and sanctioned. In many cases, the name-calling and verbal abuse from the classroom continued and multiplied on the football pitch. Very often participants would say that they “hated” football. There were two types of reflections on the football game regarding labelling. First, those who perceived themselves as more “feminine” were automatically claimed to be bad players and very often “kicked out” of the game or blamed in case of a game loss. Second, those who did not perceive themselves as physically different from the other boys usually avoided football games due to fears that they would fail to prove their masculinity on the pitch by not being able to play properly. The feeling of being different however has changed over time. While G1 and G2 tended to describe their experiences as not being able to fit into the “heterosexual market” the youngest participants, especially those born in big cities, described their early experiences with disapproval and willingness to change their school environment by taking a stand against self-objectification as “victims”. This change was brought about by two important factors: (1) the participation in LGBTQI + networks and (2) the role models from pop culture. “I wonder why the older men very often describe themselves as victims; I do not feel like a victim, I fight, I use my sexuality as an advantage rather than a disadvantage”. - Steven (23). First Encounters: Naked Bodies, Emotions and Fantasies One of the most frequently reported situations of the participants’ childhood experiences was their first physical encounters. By first physical encounters, this study understands all the situations where the participants observed another naked or partially naked male body. There were several kinds of situations reported that can be categorized as (1) reflections on older males’ bodies; (2) reflections on peers’ naked bodies and (3) reflections on the “screen bodies” – movies, comics, cartoons, toys etc. Regardless of having different experiences with different male bodies almost all of the participants in this study reflected on their memories of the first naked male bodies they had seen. Reflections on Mature Males’ Bodies Some participants report that observing neighbors provoked in them sudden physical urges and desires. Some 11 people reported this kind of physical sensation while observing male bodies from their close environment between 6 and 14 years old. ‘I must have been 7 or 8 when I watched the neighbor who was mowing the lawn and he was half-naked with a very beautiful body. I felt something very strange, then I remember I had wet dreams. I was thinking about his body very often and I did not know why, but I learnt to masturbate a bit later. Of course, I knew I could not tell anyone about this, because the other boys always discussed females and not males. - Kris (25). Similarly, Stan (35) would often dream about the man who was working in their family store. Later on, at 9 years old, he would dream about his ski teacher. He wanted “to be close to him, to touch him, to feel him”. Krasi (38) would often kiss his older cousin on the lips, he would hug him, and according to Krasi, “They all thought I was a kid, but I knew very well I feel warmth in my body when I did that”. Another most frequently reported situation where the participants remembered their interest in the male body was bathing. In some cases, especially for G1 and G2 the very first occasions of bathing with close relatives or going to a public bath that they remembered provoked “strange” physical sensations in them. Ivan (28) was 8 years old when he felt a certain type of physical desire. He used to go to public baths with his grandparents. Ivan reported remembering that when he was in the men’s pool with his grandfather, he would carefully observe male bodies, shapes and sensations. He also remembered his interest in the penis, however, he shared that he knew he should not be watching the other males’ penises – he already knew that that was a “faggot thing to do” which demonstrates early years of awareness of stigma (Fuentes, 2020) and lack of any alternative models (Cohler & Hammack, 2006; Dube, 2012). Reflections on Peers’ Naked Bodies The most frequently reported interest towards the male body involved the respondent’s peers. Very often the reflections on their first memory related to a male body were connected to the games in schools bathrooms or hidden settings on the children’s playgrounds where males “measured their penises” or “measured the length of their pees”. This appears to be a very common thing done in the childhood of almost half of the participants. The reflections on these events included realizations of desire involving self-sanctions and self-stigmatization within the “heterosexual market (McConnell-Ginet & Eckert, 2003). Angel (24) remembered being a bit ashamed of having the smallest penis among all the boys. Therefore, he would avoid participation in such masculinity reaffirming (Edwards, 2012) “competitions”, but he started fantasizing about certain boys and their physique. A very interesting ritual reported by 23 people is group masturbation. Through group masturbation, the participants understood different occasions when they would get together and masturbate. In some cases, it happened when the idea of masturbation was discovered and therefore transferred to peers. Most of the cases however were connected to a very specific phenomenon related to porn sources. Different groups of young males would gather in a place where someone would bring porn magazines which were difficult to access before the Internet, especially in smaller towns. Another occasion that would bring young boys together was whenever someone in their group would have at home one of the first video players or one of the first cable channels in town, where they would watch porn together. Those were videotapes “well-hidden” by their parents or TV1000, which played porn after 1 AM. The group masturbation appears to be one of the first very clear self-reflection of desire towards the male body. In many accounts, there was a desire to touch other penises or to have something more intimate with specific people from the group. That phenomenon was most often reported as occurring between 1995 and 2008 (G2) and those gatherings were typical for smaller towns where the children could easily get together at someone’s house. That phenomenon occurred before the mass Internet access and the socioeconomic conditions (Sediman, 2004) of the time appeared to be a prerequisite for those group gatherings. Those group gatherings were rarely reported to have occurred after 2008; however, many of the younger participants shared that they had heard about such gatherings in the past. Reflections on the “screen bodies” The wider access to the Internet allowed G2 and G3 to have access to pornography at a very young age. While G1 had access to the Internet at age of 16 on average and their computer at age of 24, G2 had access to the internet at age of 8 on average and their personal computer at age of 11. The wider access to cable channels and diverse movies also provided more opportunities to observe and compare male bodies. Some participants shared memories from a very early age. Marin (29) reflected on a memory of watching a movie with naked males taking a bath and then drawing them when he was 6 years old, which according to him puzzled his parents. Marian (22) reported remembering that at age of 7 he watched the cartoon Tarzan and he was very curious about the “muscular” body of the protagonist and he wanted to be “taken, hugged and kissed” by him. In general, the most common memories of the youngest cohort regarding first physical desires are related to gay porn. This brought an awareness of their desires and physical and emotional feelings as non-heterosexual at an earlier age compared to the generation before them. In general, the younger the participants were, the earlier they directed their first desires towards the male body. The mega-model of Lipkin (2001) defines the period before adolescence as non-sexual; however, the data from this study demonstrates that certain physical desires had been felt at a younger age. Furthermore, while G1 and G2 had experienced difficulties finding the appropriate language for their feelings and desires, G3 had struggled less to allocate their physical curiosity and desires as “normal” and “pleasurable” rather than “abnormal” and “sinful” as expressed by the two other cohorts. This discrepancy between the stage models and the field data confirmed the findings of previous studies that the rapid development of technologies within the past two decades had reorganized and intensified earlier development of sexual identity (Dube, 2000; Olive, 2012; Cohler & Hammack, 2006). Another significant change is related to the stigmatizing labels (Lipkin, 2001) and avoidance of certain people, behavior or places. Most of the participants from G3 did not express such notions and the main reason is their coming out which provides freedom of expression and resistance to certain homonormative discourses (Coleman-Fountain, 2014). “It is really stupid to avoid certain people because they are not masculine or cool, it shows your mental capacity and I think this was very typical for the older gays but they are not out which limits their whole life” - Ivo (24). Acknowledging non-heterosexual Identity and Performativity First Realization Following the mega model of Lipkin (2001), the period after the preadolescent years is considered the time for questioning one’s sexual identity. This is related to any source of information or life events that might offer a possibility for comparison of our desires and the desires of others in a socio-historical context (Coleman-Fountain, 2014; Cohler & Hammack, 2006; Dhoest, 2022). The data from this study shows that the sources of the questioning as well as the period when this process begins differ for each cohort. A very significant milestone in most of the participant’s stories was the time when they first realized that they are attracted to their sex physically and/or emotionally attaching those desires to a ”different” identity. Very typical for G1 was the romantic attachment to another boy which would reframe the social expectations for a boy to be attracted to women. On average, the time of awareness was 14 years old. The respondents from G2 reported two different reasons for realizing their own non-heterosexual identity. On the one hand, some explored the bodies in porn movies and magazines. On the other hand, some formed connections with other boys and men (mostly online) in their adolescence, which helped them realize and allocate their sexual desires in a category of “homosexual”; “gay” or “pederast” – depending on the context. The average age of awareness was 12 years. Finally, those from G3 reflected on their first awareness as an act within their environment. Those who entered adolescence after 2005 had wider access to movies, books, online forums etc., which made it possible for them not only to understand better their sexuality but also develop some identity as non-heterosexual before having any sexual encounters, which confirmed the results of other studies (Dube, 2000; Dube, 2000; Driver, 2008). The time of awareness also depends on a variety of factors however media and the Internet (Dube 2012; Martel 2018) have had a crucial role in this process. Those born in bigger cities or the capital reported an earlier year of awareness compared to those born in smaller towns. Furthermore, irrespective of the place of living, those who had a personal computer, cable channels, sex education books or supportive parents reported earlier awareness than those who grew up in homes without books, support and Internet. Performing Masculinity In Bulgaria (and not only) it is very common to ask kids from a very young age about their boyfriends or girlfriends. Often expressed as jokes, many participants reflect on these memories as “confusing” and “controlling”. “It is still going on, it started…I do not remember…since I remember myself, everyone in my town would ask me on the street when I was going to get married and have children…it is so annoying”. - Ivo (36). The memories of the first “girlfriends” among the participants dated back as far as 5 years old. In general, 19 people reported having a girlfriend when they were between 5 and 14 years old. Other 12 people reported having a girlfriend when they were between 14 and 21 years old. There are 3 different patterns of behavior and self-reflections connected to this phenomenon. For some who realized that they were attracted to men physically and emotionally when they were between 18 and 21, having a girlfriend was “the normal thing to do” reflecting on these events as inevitable. Usually, those were people who stayed detached from any non-heterosexual communities and networks (including online). “I might have had some desires towards boys, but I did not have anyone like me around and I did not know that it was a sexual desire exactly. I never spoke and compared those feelings”. - Ivan (45). For others, having a girlfriend was a strategy to avoid questions and suspicions from family and friends. In certain cases, for example in smaller Muslim towns or evangelical Roma communities, where sexual encounters before marriage were strictly forbidden, some participants used that as “an excuse” to abstain from sexual acts with their girlfriends. “I was pretty aware that I liked boys but I did not want to be called gay on the streets so I decided I would have a girlfriend” - Meto (34). Another group of people who rejected their emotions and desires towards men reflected on their relationship with girls as a way to “fix themselves”. This pattern was usually reported by people who grew up in smaller towns and religious communities. “I was so scared I would be punished by God so I promised I would have three children and fix myself, I thought it was possible” - Ivelin (47). Managing sexual identity means managing gender identity and expression (Seidman, 2004). Maintaining the expected gender roles by “having a girlfriend” was a way to “escape” (Connell, 1991; Shio & Moyer, 2021) from or deny non-heterosexual desires for many. Similarly, some individuals chose to be the penetrative (active) side in an encounter in order (according to their own words) to be “still manly” (Connell, 1992), to be “less feminine” (Edwards, 2012), to be “less sinful”. It is a pattern of behavior typical for smaller, religious communities. These events are known as pitfalls (Floyd & Stein, 2002) or setbacks (Lipkin, 2001) in the development of non-heterosexual identity. What was common in the reflections on masculinity and sexual identity was the burden of maintaining a certain masculine image and behavior which reinforced homonormativity (Hegna & Larsen, 2007). On the other hand, having a girlfriend provided a chance for comparison between intimacies with males and intimacies with females which led to a reconsideration of sexual identities and desires later in some participants’ lives. Having a girlfriend was barely reported among those from G3. In a few cases, any sexual or intimate relationship with a girl was described as a “responsible” action in an attempt to learn more about their sexuality. Many of the youngest respondents in this study had come out and respectively they did not have to conform to certain expectations of masculinity. Very often those from G3 would criticize such relationships with girls as an “egoistic” and “selfish” act where another human being (the girlfriend) is “deceived and constantly betrayed”. The dynamic in the notions of masculinity, community expectations and relationships with girls was radically changed due to the influence of the so-called youth sexual subcultures (Driver, 2008). This changing notion is usually related to the involvement with online and physical LGBTQI + community events and activities, travelling abroad and popular culture. First Sexual Encounters A central milestone in the understanding of sexual identity is connected to the physical and emotional feelings during the first sexual encounters of the participants. Very often the stories of G1 regarding their first sexual encounter are related to different spaces such as the public baths where one would experience sexual relations for the first time. The democratic changes after 1989 brought a revival in the religious institutions of different religions and denominations. On the one hand, the Bulgarian Orthodox Church established its presence in all aspects of public life, on the other hand, certain evangelical denominations and Islamic denominations entered Bulgaria and instituted certain “detraditionalization” (Darakchi, 2018) in these communities. While the participants from G1 would discuss their first sexual encounters within the context of prohibition and ‘hiding”, those from G2 often discussed their first sexual encounters within the context of the “sinfulness” of same-sex acts and the fear of HIV/AIDS. The discourses of “sin” and HIV/AIDS dominated the public conversations related to non-heterosexual people. It is not by coincidence that the biggest issues with mental health were reported for that period. The predominant HIV/AIDS discourse set another pitfall (Gillespie et al., 2021) in the development of sexual identity as some participants considered “becoming normal again”. “It was 2003 I was afraid I had caught HIV due to my first sexual act without a condom. I panicked and I started searching for information. It was all about AIDS in the newspapers. I did not have access to the Internet. I knew almost no one like me to share with… that harmed me a lot, I thought I was dying, I wanted to become normal again” - Alex (36). Some of the participants (14) had their first sexual encounters with women. One of the most decisive points regarding sexual identity formation in those participants’ lives was their first sexual encounter with men. When telling the stories of their first sexual encounters, many participants compared their sexual experiences with women to those with men. This is a common account among those from G2 who had intimate experiences with women. When making those comparisons, the participants use a very similar set of emotional words. For some, intimacy with a man made them feel “alive” for the first time. For others, the first sexual contact with a man compared to the one with a woman gave them an answer to the question of why they felt different in the past. For some others, sexual contact with men felt “natural” for the first time. It is hard to summarize the words used in those accounts; however, the male body compared to the female body was described as “tempting,” “attractive,” “with the perfect smell,” “natural,” “real,” “unique” and others. In conclusion, having first sexual acts with women was rarely reported among G3. The cultural and social liberation and access to resources allowed many to reflect on their sexual desires earlier in a process of subjectification (Dowsett, 2015) within an “alternative” and liberating language framework” (Gillespie et al., 2021). In these specific cases, very often that was an act of self-reflection, of seeking to understand more about their sexuality. What was significantly different for the three generations in the development of sexual identity in the adolescent years was the changes in the available categories depicting sexual belonging, desires and preferences within a “body liberation” (Rothmann, 2013) framework. Identity categories such as sapiosexual, omnisexual, pansexual, queer and others usually did not exist in the vocabulary of G1 and G2 when they reflected on their experiences. The diversification of the identifications and the awareness of alternative “queer” identity categories have been reported in many studies among the youngest cohorts( Cohler & Hammack 2006; Gillespie et al., 2021; Dhoest, 2022) and this study confirms these findings. The LGBTQI + activism in Bulgaria, as well as YouTube videos and Facebook groups, made an increasing number of terms available to young people even in their preadolescent years, which resulted in a better allocation of sexual desires in different categories and according to the participants that helped them to “find themselves”. “YouTube, I got to know everything from YouTube, there are many vloggers who taught me at an early age and it helped me a lot not to wonder what was happening to me” - Milen (21). Sources of Information The respondents paid significant attention to the self-acceptance of their sexual identity or the “coming out to yourself” as defined by one of the participants. There were strong generational and socioeconomic differences in the self-acceptance of non-heterosexual desires and emotions. The “non-heterosexual” places appear to have played a big role in the individuals’ understanding of sexuality and difference. Before 1989, the main places where the respondents got to know other non-heterosexual males were public baths. Located in the bigger cities and the capital, the public baths provided one of the first opportunities for the participants to interact with other individuals, share their experiences and reflect on their sexual desires. In Sofia, another opportunity used to be Culture (Kultura) cinema where some of the participants met other males and initiated intimate contact for the first time. “Toploto beach” near Varna happened to be the place where Dani (67) used to go every summer and meet not only Bulgarian gay friends but also many tourists from Poland, Hungary and Czechoslovakia. While some of the participants had the chance to travel abroad and to reflect on their desires in a foreign setting even before 1989, the years following the fall of communism provided for many more the chance to travel to different “Western countries,“ which appears to be one of the strongest influences on the developments of non-heterosexual identity for G1 and G2. These visits were very often connected with different conferences, training or tourism. After 2005 many people would go for the first time in their life to a gay bar in Sofia, engage with an LGBTQI + organization or join online forums and dating apps (Darakchi, 2021). In the years after 2010, the interactions took place predominantly in online settings. Most of the bars in Sofia ceased to exist and those in Varna and Plovdiv closed their doors as well. The Internet and dating pages used to be the main places for interactions. In 2021 the STEPS, an LGBTQI + community space, opened its doors and started playing a significant role as a safe space for exhibitions, concerts, training and community meetings. These dynamics of the community spaces also created different attitudes towards self-acceptance among the participants. While the gay scene in Sofia before 1989 provided opportunities for face-to-face meetings and reconsideration of sexual identity, the period after 2010 was connected to a certain “anonymization” (Mowlabocus, 2016) among some of the participants with limited opportunities for interactions, especially among those who search “only sexual encounters”. This, however, should not be considered as a setback because some who live in smaller places or disengage from the scene or the mainstream non-heterosexual community have easier access to sexual encounters strategically revealing their identities which often resulted in friendships. Self-acceptance strongly depends on the sources of information and interaction. These sources are connected to examples and cases where sex is not isolated as a physical desire but is celebrated as a feature of the individual (Seidman, 2004) and self. Before 1989, the occasional encounters with foreigners in Bulgaria and the travels to other countries provided opportunities for some to reflect on their sexual desires and experiences by watching movies and reading books. Some of the participants had a chance to travel to Budapest, Prague Warsaw or Berlin bringing back ideas of sexual identity. The books that existed in some family libraries and played a role in self-acceptance before 1989 were “Men and Women Intimately” (1967) by Siegfried Schnabl and “The Portrait of Dorian Gray” by Oscar Wilde. The movies which provoked self-reflection used to be “Brideshead Revisited” (1967) first released in Bulgaria in 1985 and “Death in Venice” (1971) by Thomas Mann. The music icons of that period were Madame Dаlida (1933–1987) and Lili Ivanova. The years between 1995 and 2010 provided several opportunities for interactions and self-reflection. That period also marked wider access to the Internet, web pages and forums for non-heterosexual people. The book “Men and Women Intimately” (1967) continued to play a role in the self-understanding of the participants growing up in this period and it has been mentioned up to 2006. Very important sources for self-reflection became Stalik blog and the dating webpages and forums on momcheto.com, elmaz.bg and planetromeo.com. Notable movies and series which played role in the participants’ self-reflections in that period were Will and Grace (1998), The Next Best Thing (2000), and Queer as Folk (1999), which provided not only role models for many but the language and the categories describing identities and preferences. Desperate Housewives (2004), Brokeback Mountain (2005), Sex and the City (1998) and many others became accessible for pirate downloading or were offered on cable channels. When it comes to music one of the most mentioned names which inspired many and became role models were these of Freddy Mercury, Madonna, George Michael and Elton John. Despite the increasing availability of sources of information and role models, self-acceptance during that period was negotiated between those new sources of information and interactions and the dominating discourse in the media, which usually connected non-heterosexuality with HIV/AIDS and depicted LGBTQI + organizations as a threat to the birth rates and the national identity. After 2010, the most influential movies and series which provided an opportunity to the participants who grew up during this period to question and self-objectify themselves were The New Neighbors (2007), Physics or Chemistry (2008) and Glee (2009), which became widely available on cable channels and even national TV channels. The expansion of Netflix and HBO go in recent years provided a variety of LGBTQI + movies and series which have influenced the understanding of sexuality and non-heterosexuality. Notable names from this period are Pride (2014) Sense8 (2015), When We Rise (2017), Elite (2018), Euphoria (2019), Sex Education (2019) and other politically engaged series and movies. The movie Call Me by Your Name (2017) and the book on which it was based attracted significant attention. The pop icons of this period are Lady Gaga, Billy Eilis, Sam Smith and Olly Alexander. Although this is not an exhaustive list of sources that have contributed to the understanding of non-heterosexual identity, these resources played a significant role in the reflections of the participants. In general, these resources predefined the understanding of “good” and “bad” sexuality and respectively sexual identity (Seidman, 2004). In general, the available sources of information for G1 and G2 mediated processes of sexual orientation and identity objectification and self-objectification (Dowsett, 2015) where the non-heterosexual desires have been gradually accommodated into identity and the self-stigmatization (Dube, 2000; Driver 2008) has significantly decreased. The first two generations are more likely to identify with a non-heterosexual identity ( gay, homosexual, pederast ( in a joking manner) which promotes monogamy and good “citizenship” (Seidman, 2004) while many from G3 have subjectified their identities (Dowsett, 2015) and are more likely to reflect on the heteronormative and homonormative structures and consider their sexuality in terms of sex-positivity, where sex is celebrated and open relationships are viewed as an alternative to the monogamous couples. Another significant trend is the understanding of non-heterosexuality as natural. Often citing the famous “Born This Way” by Lady Gaga, some from G3 have a very distinct approach towards their sexualities compared to G1 and G2. The sources of information after 2010 have provided a very positive image of non-heterosexuality which is often seen as “natural”, “diverse” and “inclusive” of psychical and mental differences by G3. It is a process that Lipkin (2001) calls “engagement in a critique of conventional heterosexism”. Of course, this does not mean that all the participants from G3 have these notions of their sexualities. On the contrary, there were a few people from G3 who would not have these notions and they were usually very detached from the LGBTQI + community networks and online forums. Rephrasing Lipkin’s (2001) words, these participants avoided having any “dimensions of meaning” to their sexual desires although they belonged to a community sexually, mostly online. Regarding the sources of information, the data suggests that the wider the individual’s participation in community networks and engagement with popular culture is, the earlier self-acceptance is observed. Those who stayed out of cultural events, online movies and book forums usually expressed negativism towards the popular culture and the models of behavior. These people tend to disassociate their physical desires from their public life, behaviors and interests. The greater the involvement in the LGBTQI + culture and networks, the more inclusive approach towards sex as a positive, intimate and romantic experience is expressed. The self-acceptance of the participants also depended on their families and the general attitudes towards difference and non-heterosexual issues. Despite certain generational differences, the data suggests that the higher educational status of the parents provided more opportunities for self-acceptance. In most cases, those opportunities were connected with providing appropriate literature and support. In some cases, that was done by providing books on similar topics. In other cases, it was done by providing appropriate occasions for conversations. Regarding gender differences, the mother was the parental figure who initiated the conversation and in many cases, the mother managed the father’s fears and prejudices. In general, the younger the parents were the more reflexive approach they had. The generational aspect was significantly connected with the notion of non-heterosexual people who were already a part of the parents’ personal and working networks. The data from this study also suggests that self-acceptance is a continuous process with pitfalls and setbacks which confirms the life-long nature of sexual identity formation (D’Augelli 1994). It is a process of “continuous upgrading,“ as defined by Kris who is 54 years old, however, he has been considering notions and ideas of three generations and he is willing to engage with different discourses and ideas challenging his stereotypes and ideas of being non-heterosexual. Although marked as a stage in the stage models, the self-reflection and the self-acceptance is a long and in some cases reversible process. This is the case of Alex (39) and Kiril (38). Alex, who was born in a small town, used to consider same-sex relations and sexual practices as sinful. Alex moved to the capital where he had the opportunity to interact with many other non-heterosexual people and to travel to many countries. His self-acceptance has been a process of learning and interacting; however, he still self-justifies certain desires and behavior in a continuous fight with his initial views of same-sex practices as “dirty, sinful, and forbidden”. Self-acceptance in this case can also shift in different settings. “When I spend a month in my home town, I start feeling more sinful, it is all about religion and punishments there, but when I move back to my life in Sofia I feel more relieved seeing how normal it all is” - Georgi (39). On the opposite side is Kiril who used to hang out with Alex during their student years and they used to share similar ideas and dreams. Kiril stayed in the smaller town and got a job. He does not have gay networks; in his own words, he is not happy with his non-heterosexuality, he is outside of any LGBTQI + networks. His understanding of life and intimacy is very negative and blameful. He considers homosexuality as abnormal and deviant and “does not see a point in it all”. Discussion The stage models categorize preadolescence as a non-sexual period. However, the data from this study, building on similar research (Olive, 2012; Giano, 2019), has proven that this does not correspond to the individual experiences, especially during the past decade. Initializing these early signs of one’s non-heterosexual awareness is important for the scholarship on non-heterosexual identities and sexual development models. It would not only improve the research methodology and the reconsideration of the sexual development models but it can also provide evidence for more effective public policies and prevention programs. The early years experiences were better understood, reflected and self-subjectified at a younger age among those who grew up during the past decade compared to the previous generations due to the advancement of the Internet, video streaming platforms, mobility and LGBTQI + community involvement. Those from G1 and G2 self-labeled their same-sex attraction mainly through a “sex-centred” sequence (Dube, 2000), after engaging in sexual activities while the majority of the respondents from G3 self-labeled their same-sex attraction mainly through an “identity-centred” sequence (Dube, 2000) before engaging in sexual activities.  However, these results do not fully correspond to what is categorized as “global gay culture” (Martel, 2018) because almost half of the respondents mainly from G1 and G2 do not recognize the terms gay and gay identity as relevant to them. This poses a challenge for further research that “celebrates” the progressive advancement of “gay rights” and might overlook criticism of “gay” as a cultural identity, missing out on misrecognitions of widely used sexual identity categories. The youngest generation in the study is less likely to have suffered from psychological discomfort, confusion and intimate life difficulties compared to the generations that grew up before them. The youngest generation has also shared a more positive image of themselves as being non-heterosexual. This confirms previous studies (Giano, 2019; Fuentes, 2020) dealing with the effect of early self-awareness and acceptance of non-heterosexual people. However, in the Bulgarian context, there is no single state school program for inclusion and support of non-heterosexual people and the attempt for such a discussion has been overshadowed and sabotaged by the anti-gender mobilization (Darakchi, 2019). The first sexual encounters were reported to be one of the main milestones in the development of sexual identity during adolescence. This was even more explicit among those who had had previous sexual experiences with women, which allowed them to compare their physical and emotional feelings when engaging sexually with men and with women. While the age of the first sexual encounter has been dropping within the last 15 years, sexual encounters with females were less often reported and in certain cases, this was qualified by the youngest participants as a selfish act connected to injustice towards the women who were involved in this type of relationship. These results reconfirm the conclusions from other studies (Giano, 2019, Gillespie et al., 2021) and this requires more attention on the public policies for the prevention of STIs and ethical sexual conduct. The formation of sexual identity is directly influenced by income, place of living and community involvement. While family income and the place of living played a bigger role in the participants’ sexual awareness for G1 and G2, the rapid technological changes have allowed for literally everyone from G3 to have early access to information and materials, which not only have a positive effect on their self-stigmatization and wellbeing but also challenges the scholarship which defines LGBTQI + movement and culture as explicitly class defined structure (Barrett & Pollack, 2005). The recent “anti-gender campaigns” in Bulgaria have motivated some of the participants from G3 to join politically engaged community networks and actions. Although this phenomenon is quite recent it has strengthened the earlier self-awareness of some further motivating political participation. These results are quite contrary to what Ghaziani (2011) describes as a “post-gay” era where youngsters prefer mixed networks to non-heterosexual communities. The same trend has been recently confirmed by Dhoest (2022) who explored generational differences in Flanders, Belgium. Ghaziani’s (2011) study is carried out before 2011 when the “anti-gender campaigns” were not as organized and visible as today (Kuhar & Paternotte, 2017) however in the case of Dhoest (2022) this tendency is confirmed. This might mean that either the influence of the “anti-gender campaigns” is not as strong as in the Bulgarian context or the influence of these movements might not have been included in the research focus, especially in the latter study. Finally, the results from this study can serve as a ground theory for “provincialization” of the sexual development models in a post-communist context with specific generational notions which differ significantly from those described in the “Western” scholarship. On the other hand, the data suggest that the awareness of one’s non-heterosexual orientation and identity develops regardless of the political context and the “nature” of non-heterosexual desires and identities in Bulgaria was a “local process” long before the “global gay culture” (Martel, 2018) formed during the recent decades. Author Contributions This article has no co-authors. Funding This work is funded by Research Foundation - Flanders (FWO) under the Marie Skłodowska-Curie Actions - Seal of Excellence Postdoctoral Fellowship. Statements and Declarations Competing Interests The author has no relevant financial or non-financial interests to disclose. Ethics Approval This study was performed in line with the principles of the Declaration of Helsinki. Approval was granted by the Ethics Committee for the Social Sciences and Humanities of the University of Antwerp issued on 16.03.2020/Reference number: SHW_20_09. Consent to Participate Informed consent was obtained from all individual participants included in the study. Consent to Publish The author affirms that human research participants provided informed consent for the publication of the data they shared in a summarized format. Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References Barrett DC Pollack LM Whose gay community? Social class, sexual self-expression, and gay community involvement Sociological Quarterly 2005 46 3 437 456 10.1111/j.1533-8525.2005.00021.x Bereket T Adam B The emergence of gay identities in contemporary Turkey Sexualities 2006 9 2 131 151 10.1177/1363460706063116 Carless D Douglas K Narrative research The Journal of Positive Psychology: dedicated to furthering research and promoting good practice 2017 12 3 307 308 10.1080/17439760.2016.1262611 Chauncey, G. (2008). Gay New York: Gender, urban culture, and the making of the gay male world, 1890–1940. Hachette UK. Cohler, B. J., & Hammack, P. L. (2006). Making a Gay Identity: Life Story and the construction of a coherent self. American Psychological Association. Coleman-Fountain, E. (2014). Understanding narrative identity through lesbian and gay youth. Springer. Connell R A very straight gay: masculinity, homosexual experience, and the dynamics of gender American Sociological Review 1992 57 6 735 10.2307/2096120 Darakchi S Gender, religion, and identity: modernization of gender roles among the bulgarian Muslims (Pomaks) Women’s Studies International Forum 2018 70 1 8 10.1016/j.wsif.2018.07.002 Darakchi S “The western feminists want to make us gay”: Nationalism, heteronormativity, and violence against women in Bulgaria in times of “anti-gender campaigns” Sexuality & Culture 2019 23 4 1208 1229 10.1007/s12119-019-09611-9 Darakchi, S. (2021). Bulgarian LGBTQI + movement: Generations, identifications, and tendencies. Sexualities, 13634607211000201. D’Augelli, A. R. (1994). Identity development and sexual orientation: toward a model of lesbian, gay, and bisexual development. In E. J. Trickett, R. J. Watts, & D. Birman (Eds.), Human diversity: perspectives on people in context (pp. 312–333). Jossey-Bass/Wiley. Dhoest, A. (2022). Generations and shifting sexual identifications among flemish non-straight men. Journal of Homosexuality, 1–24. DOI: 10.1080/00918369.2022.2048164. Dowsett GW Practicing desire: homosexual sex in the era of AIDS 1996 California Stanford University Press Dowsett, G. (2015). ‘And next, just for your enjoyment!’: sex, technology and the constitution of desire. Culture, Health & Sexuality, 17(4), 527–539. Driver, S. (2008). Queer youth cultures. SUNY Press. Dube EM The role of sexual behavior in the identification process of gay and bisexual males Journal of Sex Research 2000 37 2 123 132 10.1080/00224490009552029 Duberman MB Reclaiming the gay past Reviews in American History 1988 16 4 515 525 10.2307/2702347 Edwards, T. (2012). Erotics and politics: Gay male sexuality, masculinity and feminism. Routledge. Eguchi S Negotiating hegemonic masculinity: the rhetorical strategy of “straight-acting” among gay men Journal of Intercultural Communication Research 2009 38 3 193 209 10.1080/17475759.2009.508892 Floyd FJ Stein TS Sexual orientation identity formation among gay, lesbian, and bisexual youths: multiple patterns of milestone experiences Journal of Research on Adolescence 2002 12 2 167 191 10.1111/1532-7795.00030 Fuentes, A. (2020). The development of gay college men in traditional fraternities.Journal of the Student Personnel Association at Indiana University,68–77. Giano Z The influence of online experiences: the shaping of gay male identities Journal of Homosexuality 2019 68 5 872 886 10.1080/00918369.2019.1667159 31532331 Gillespie I Armstrong H Ingham R Exploring reflections, motivations, and experiential outcomes of first same-sex/gender sexual experiences among lesbian, gay, bisexual, and other sexual minority individuals The Journal of Sex Research 2021 59 1 26 38 10.1080/00224499.2021.1960944 34406093 Harry J Sampling gay men Journal of Sex Research 1986 22 1 21 34 10.1080/00224498609551287 Hegna K Larsen C Straightening out the queer? Same-sex experience and attraction among young people in Norway Culture Health & Sexuality 2007 9 1 15 30 10.1080/13691050600965984 Husu H Bourdieu and social movements: considering identity movements in terms of field, capital and habitus Social Movement Studies 2013 12 3 264 279 10.1080/14742837.2012.704174 Klinkenberg D Rose S Dating scripts of gay men and lesbians Journal of homosexuality 1994 26 4 23 35 10.1300/J082v26n04_02 8006419 Korolczuk E Graff A Gender as “Ebola from Brussels”: the anticolonial frame and the rise of illiberal populism Signs: Journal of Women in Culture and Society 2018 43 4 797 821 10.1086/696691 Kuhar, R., & Paternotte, D. (Eds.). (2017). Anti-gender campaigns in Europe: mobilizing against equality. Rowman & Littlefield. Lipkin, A. (2001). Understanding homosexuality, changing schools. Routledge. Martel, F. (2018). Global gay: how gay culture is changing the world. MIT Press. McConnell-Ginet, S., & Eckert, P. (2003). Language and gender. Cambridge University Press. Merriam SB Qualitative research in practice: examples for discussion and analysis 2002 San Francisco, CA Jossey-Bass Mowlabocus, S. (2016). Gaydar culture: Gay men, technology and embodiment in the digital age. Routledge. Olive JL Reflections on the life histories of today’s LGBQ postsecondary students Journal of LGBT Youth 2012 9 3 247 265 10.1080/19361653.2012.652854 Parker R Beneath the Equator: cultures of desire, male homosexuality and emerging gay communities in Brazil 1989 NY Routledge Rothmann J ‘Doing’ and ‘using’ sexual orientation: the role of gay male pornographic film in the identity construction of gay men South African Review of Sociology 2013 44 3 22 41 10.1080/21528586.2013.817048 Seidman, S. (2004). Beyond the closet: the transformation of gay and lesbian life. Psychology Press. Shio J Moyer E Navigating norms of masculinity: tactical gender performances among gay men in Dar Es Salaam, Tanzania Gender Place & Culture 2021 28 6 853 869 10.1080/0966369X.2020.1759513

0

PMC9754303

NO-CC CODE

2022-12-16 23:26:35

no

Sex Cult. 2022 Dec 15;:1-21

utf-8

Sex Cult

10.1007/s12119-022-10049-9

oa_other

==== Front Neural Comput Appl Neural Comput Appl Neural Computing & Applications 0941-0643 1433-3058 Springer London London 8127 10.1007/s00521-022-08127-y Original Article A residual network-based framework for COVID-19 detection from CXR images Kibriya Hareem hareem.kibriya@students.uettaxila.edu.pk 1 http://orcid.org/0000-0002-3143-689X Amin Rashid rashid.amin@uettaxila.edu.pk 12 1 grid.442854.b Department of Computer Sciences, University of Engineering and Technology, Taxila, Pakistan 2 Department of Computer Science, University of Chakwal, Chakwal, 48800, Pakistan 15 12 2022 112 17 11 2021 28 11 2022 © The Author(s), under exclusive licence to Springer-Verlag London Ltd., part of Springer Nature 2022, Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. In late 2019, a new Coronavirus disease (COVID-19) appeared in Wuhan, Hubei Province, China. The virus began to spread throughout many countries, affecting a large population. Polymerase chain reaction is currently being utilized to diagnose COVID-19 in suspected patients; however, its sensitivity is quite low. The researchers also developed automated approaches for reliably and timely identifying COVID-19 from X-ray images. However, traditional machine learning-based image classification algorithms necessitate manual image segmentation and feature extraction, which is a time-consuming task. Due to promising results and robust performance, Convolutional Neural Network (CNN)-based techniques are being used widely to classify COVID-19 from Chest X-rays (CXR). This study explores CNN-based COVID-19 classification methods. A series of experiments aimed at COVID-19 detection and classification validates the viability of our proposed framework. Initially, the dataset is preprocessed and then fed into two Residual Network (ResNet) architectures for deep feature extraction, such as ResNet18 and ResNet50, whereas support vector machines with its multiple kernels, including Quadratic, Linear, Gaussian and Cubic, are used to classify these features. The experimental results suggest that the proposed framework efficiently detects COVID-19 from CXR images. The proposed framework obtained the best accuracy of 97.3% using ResNet50. Keywords COVID-19 Machine learning CXR images SVM ResNet50 ==== Body pmcIntroduction Coronaviruses have been responsible for two large-scale outbreaks in the last two decades: SARS and MERS. COVID-19, also known as SARS-CoV-2, is a novel human coronavirus that arose in Wuhan in Hubei Province, China, in late 2019 [1]. This virus has also been reported in almost all continents, including Europe, Australia, Asia, and America [2]. Until the 16th of February, 2022, 415 million confirmed cases had been reported worldwide, with a total of 5.84 million deaths [3]. A study shows the zoonotic origin of COVID-19, and the fast expansion suggests continuous person-to-person transmission [4]. COVID-19 is a respiratory disease characterized by symptoms ranging from a moderate infection in upper respiratory system to acute respiratory distress syndrome (ARDS) [5]. COVID-19 symptoms might differ from person to person, but some are typical. These symptoms typically emerge between 3 and 14 days. The Center for Disease Control and Prevention (CDC) [6] lists major COVID-19 symptoms: dry cough, trouble breathing, fever, muscle discomfort, sore throat, loss of smell and taste. Some of the less common symptoms are nausea, vomiting, and diarrhea. This virus has been shown to be airborne for up to 3 h, 4 h on copper surfaces, and nearly 72 h on plastic and stainless materials. However, identifying the actual nature of the virus remains an unresolved topic in the medical research world [7]. Due to surge in cases, the World Health Organization (WHO) declared COVID-19 a pandemic and public health emergency of worldwide significance [8, 9]. Early and precise diagnosis of acute COVID-19 is necessary to reduce healthcare capacity and overall death rate [10]. Since the emergence of infections, the world governments have put plenty of countermeasures to mitigate the devastation being caused. Researchers and health organizations all over the world are working feverishly to develop early disease detection and treatment approaches [11]. Scientists from all around the world have developed various COVID-19 vaccines. Many of these vaccines have already been approved and are being used for immunization in numerous countries worldwide. These vaccines include Pfizer-BioNTech, Moderna, Oxford-AstraZeneca, Sinovac and Sputnik V. Despite widespread vaccination campaigns, coronavirus infections have risen, owing primarily to new variants. However, in many countries, mass vaccine adaptation remains a problem of public health logistics (e.g., manufacturing, storing, and distributing the vaccine and mass immunization) and leadership. The problem is ascribed to various factors, including vaccination resistance among citizens and vaccine nationalism [12, 13]. Further study is required to address outstanding questions such as: would these vaccines be able to manage the pandemic? What is the efficacy of these vaccines against developing various COVID-19 variants? What are the adverse effects of contemporary vaccines, studied, developed, and tested at a fast pace on various population statistics? [7] RT-PCR tests must validate the diagnosis of COVID-19 according to the recommendations by the Chinese government [14]. However, it has some limitations, such as long delays in obtaining the results and low sensitivity. The patients with high clinical suspicion of the virus test falsely negative on initial RT-PCR tests require multiple tests to validate the results [15]. Low test sensitivity may occur due to the following factors: suboptimal clinical sample procedures, fluctuations in viral load, and manufacturer test kit sensitivity. With communities experiencing an increase in patients, negative RT-PCR test is becoming very difficult to handle [16]. According to October 2020 recommendations by WHO, chest imaging evaluation is an effective method for detecting clinical signs of COVID-19 in patients because the infection is related to the pulmonary system [17]. The study presented by Ai et al. [14] compares the sensitivity of the Chest computed tomography (CT) and the RT-PCR tests. The Chest CT obtained a better sensitivity score than RT-PCR for detecting COVID-19 in suspected patients. Other imaging modalities include X-rays, MRIs, and needle biopsy of the lung. Compared to the CT Scans, Chest X-ray is the most commonly utilized to identify COVID-19 because CT imaging is expensive, and the high ionizing radiation levels may pose health hazards to pregnant women and children. Moreover, it takes longer to scan image and is scarcely available in low-income countries [18]. On the other hand, X-ray imaging is widely used in many medical and epidemiological applications due to its widespread availability [19, 20]. Due to ease of use by radiologists, lightweight operating speed and cheaper cost, the chest X-ray is a well-suited imaging modality for diagnosing pulmonary infections [21]. Over the past two years, various frameworks employing CNNs have been presented to detect COVID-19 from chest X-Ray images. CNN is a type of Deep Neural Networks (DNN) inspired by the biological working of the human brain and used to analyze visual imagery. CNNs are extremely efficient and require little or no preprocessing before classification. These architectures automatically extract deep features and perform classification, whereas typical Machine Learning (ML) algorithms require a manual segmentation and handcrafted feature extraction. This independence of CNNs from human interaction and prior information is a significant asset. Doctors can benefit from automated disease classification systems during the treatment procedure [22]. Several studies have been proposed to detect COVID-19 and other lung illnesses using Chest CT and X-rays; however, these techniques have limited classification accuracy and a large false-positive rate due to poor data quality and inefficient methods. Hence, there is a need for a robust approach that can distinguish between COVID-19 and normal images efficiently. As a result, we present a novel COVID-19 identification and classification approach based on well-known ResNet architectures for deep feature extraction and SVM for image classification. The contributions of the paper are as follows:The input CXR images are initially preprocessed to enhance the essential features helpful for classification. We improved the contrast of dataset images and resized them to fit the CNNs’ input layer size. The deep CNN features are extracted using final Fully Connected layers from Residual Network architectures such as ResNet18 and ResNet50. The SVM classifier with four kernel functions such as Quadratic, Linear, Gaussian, and Cubic has been used to classify the deep CNN features. The proposed method is compared to existing state-of-the-art COVID-19 classification and identification models to determine its robustness. The performance of the suggested method is evaluated using various measures such as recall, precision, f1-score, and accuracy. The proposed technique is also validated on a cross-dataset validation scenario. Related work Artificial intelligence and DL-based new methodologies have significantly impacted medical image analysis, particularly in disease detection. Since the emergence of COVID-19, researchers have been developing DL and ML based techniques for early diagnosis of COVID-19 using Chest CTs and X-rays. However, accurate diagnosis is still a challenge due to the lack of quality data and robust techniques. Xu et al. [23] provided an early prediction approach for identifying COVID-19 using ResNet18 that obtained 86.7% accuracy. Hemdan et al. [24] deployed a pre-trained CNN model, i.e., VGG16, to discriminate between COVID-19 and Normal images with 90% accuracy. Singh et al. [25] presented a novel multi-objective differential evolution-based CNN architecture for distinguishing COVID-19 from normal CT images. To detect COVID-19 from images, Loey et al. [26] used pre-trained CNNs such as GoogLeNet, AlexNet, and ResNet18 combined with Generative Adversarial Networks (GAN). Hernandez et al. [27] suggested a COVID-19 detection and classification system using fine-tuned CNN architectures, including ResNet, VGG, and DenseNet, and reported an overall accuracy of 90%. COVID-Net, a 19-layered CNN architecture based on the principle of Residual Networks, was created by Wang et al. [20] to distinguish between normal images, bacterial infection images, and non-COVID-19 viral, and COVID-19 viral infection images. COVID-Net provided the model with the highest classification accuracy of 93.3%. In another study, Ismael et al. [28] developed an end-to-end CNN model composed of 21 layers to classify COVID-19. The novel CNN architecture achieved 91.5% classification accuracy. Despite creating a modified CNN, the system's overall accuracy was low. In another study, Narin et al. [29] proposed a ResNet50, Inception-ResNet V2 and Inception-V3 based approach to classify COVID-19. However, in order to generalize the approach to real-world scenarios, it must be trained and evaluated on datasets with more image samples. Wang et al. [30] employed pre-trained CNN architectures to discriminate between COVID-19 and healthy images. The study reported 89.5% classification accuracy. Pathak et al. [31] employed various transfer learned residual networks architectures to classify COVID-19 infected images. They obtained the highest accuracy of 93% on publically available CT image datasets. However, the system is not robust in classifying high-intensity variations noisy images. Zhao et al. [32] applied transfer learned DenseNet architecture with 169 layers to identify the presence of COVID-19 from CT images. The system achieved 84.7% accuracy. However, the system is computationally exhausting. The authors in Ibrahim et al. [18] fine-tuned AlexNet to identify and classify COVID-19 from images, but they did not equate the performance of their proposed system to current approaches. On the other hand, Bai et al. [33] suggest a COVID-19 detection approach employing Multi-layered perceptron with LSTM. However, the system is resource-intensive due to the combination of LSTM and multi-layered perceptron. Table 1 provides a summary of the existing COVID-19 identification methods, as well as their limitations.Table 1 Summary of existing COVID-19 identification systems Reference DL model Limitations Xu et al. [23] ResNet18 The technique needs to be trained and evaluated on a larger dataset to generalize better on real scenarios Wang et al. [30] Inception The work is not robust to images with noise and intensity variation Hernandez et al. [27] ResNet, VGG, DenseNet The method needs performance improvement Hemdan et al. [24] VGG16 They used a very limited dataset with only 100 images. Hence, the technique needs to be validated Singh et al.[25] CNN The model needs to be trained and evaluated on a large database Zhao et al. [32] DenseNet The network uses a lot of memory thus is computationally expensive Wang et al. [20] CNN The technique may not be able to classify images with intensity variations Pathak et al. [31] Residual Networks The work is not robust to the classification of images with noise and contrasting intensity values Bai et al. [33] Multi-layered perceptron with LSTM Combining LSTM and multi-layered perceptron increases the parameters; thus the inference time will be high Ismael et al. [28] CNN, ResNet, VGG16, 19 The technique may fail to perform on images with intensity variation. Moreover, it needs a larger database for system training and evaluation Loey et al. [26] GoogLeNet, AlexNet and ResNet18, GAN They used a very small dataset containing only 306 samples for four classes, i.e., normal, pneumonia (viral and bacterial) and COVID-19 Ibrahim et al. [18] AlexNet The proposed method's performance is not compared to state-of-the-art techniques Narin et al. [29] Inception-V3, Inception-ResNet V2, ResNet50 The method may not perform well in real-time because it is trained and evaluated in a very small dataset Proposed methodology In this paper, we propose a DL based technique to detect the presence of COVID-19 in CXR images. The detailed block diagram of the proposed method is shown in Fig. 1. Initially, the dataset images are preprocessed before being fed into two Residual Network models for deep CNN feature extraction, such as ResNet18 or ResNet50. The deep feature vectors are classified using SVM with different kernel functions: Linear, Gaussian, Cubic, and Quadratic.Data preprocessing Fig. 1 Diagram illustrating the proposed framework This section discusses the preprocessing techniques applied to dataset images before feeding them to CNNs for feature extraction. Initially, the contrast of CXR images in dataset is enhanced by mapping the intensity values of an input image to new values after saturating the top and bottom 1% of the pixel values. Figure 2 depicts a CXR image before and after contrast enhancement. Contrast enhancement is a technique that alters the pixel intensity of an image to use as many bins as possible. It distinguishes between dark and bright areas in an image and eliminates any ambiguity that might otherwise exist between different image regions [34]. The images are then resized to match the input layer size of CNNs. Deep feature extraction using residual networks Fig. 2 Image before and after contrast enhancement CNNs are widely used in various research fields due to their superior performance over traditional ML-based techniques. The architecture of CNN is biologically inspired by the human brain area called the visual cortex that contains cells sensitive to visual perception. The use of CNNs in ML derives from a 1962 study by Hubel and Weisel, who discovered that some individual neuron cells in the brain that activate only in the presence of specific edges and shapes [35]. Usually, the CNN architectures have numerous layers, as shown in Fig. 3 (convolution, pooling and fully connected) stacked on top of each other. These models learn more complicated features with an increase in architecture’s depth. The convolution layer extract features from an image by computing the dot product between the image and the kernel. The kernel slides through the image region to compute the features, bypassing particular pixels called stride. As first described by Marr, the filter detects various low-level features such as colors, edges, curves, boundaries, virtual lines and high-level features such as local surfaces, orientations, and discontinuities [36]. The convolved portion of the image is called a receptive field. The features extracted during this process serve as input to deep layers in the architecture. This layer is followed by an activation function that is responsible for converting the node's summed weighted input into the node's activation for that input. Due to ease in training and better performance, ReLU has become a default activation function for neural networks. It is responsible for reducing all negative values to zero [37].Fig. 3 General architecture of the CNN One of the limitations of convolution layer’s feature map output is that it records the precise position details of the features in an input. Thus, very minor changes in the position of the feature in an image can result in a distinct feature map. This can occur when the input image is rotated, re-cropped, shifted, etc. Hence, downsampling is frequently used to create a reduced resolution version of an input signal that maintains structural parts of an image but lacks very fine or minor details that may be invaluable for the task. This downsampling process is performed using a Pooling layer that is usually placed after a Convolution layer. It decreases the spatial dimensions of an image following the convolution process. If we apply the Fully Connected Layer straight after the Convolution Layer, the system may become computationally expensive due to the high dimensionality of the image. As a result, pooling layers are utilized to scale the images and lower the dataset's spatial volume [38]. The output obtained from the convolution layers usually represents high-level features in the data. This output is generally flattened and connected to the output layer by adding a fully-connected layer to learn nonlinear combinations of these features. All neurons in a fully connected layer connect to all neurons in the previous layer. This layer combines all of the local features gathered by the preceding layers across the image to discover bigger patterns. The classifier then performs classification based on these features [39, 40]. Over the past few years, remarkable development has been made in Deep Learning. Following CNN architecture AlexNet, which won the ImageNet 2012 competition, each successive winning architecture employs more layers in a deep neural network to lower error rates. Deep Neural Networks are becoming more complex and deeper with time. It has been demonstrated that increasing the number of layers in these networks increases its performance and robustness for image-related tasks. Hence, the researchers increase the depth of the neural networks to extract important features from complex images. As a result, the initial layers may identify edges, and succeeding layers may detect complete objects at the end. However, this works for networks with fewer layers, but with network convergence, the accuracy of these models tends to be saturated. This decrease in the accuracy score is not due to overfitting but rather to a rise in error rate after increasing the depth of the network, as current solvers are unable to optimize the network efficiently. This runs counter to the popular belief that increasing layers in the network improves its performance. It may be noted that this problem does not appear due to network overfitting because, in such scenario, the dropout and regularization techniques can be used to solve this problem. But it is mostly present because of the well-known Vanishing/Exploding gradient problem. This problem causes the gradients of the loss function to approach zero or very large when more layers with specific activation functions are added to neural networks, thus making the network difficult to train and increasing train and test errors [41, 42]. In 2015, the researchers at Microsoft proposed Residual Network (ResNet). The introduction of ResNets aided in the training of very deep networks. ResNets use shortcut connections to skip one or more layers, as shown in Fig. 4. The layers fit a residual mapping rather than assuming each stack of layers immediately fits a desired underlying mapping. The original mapping is transformed into F(x) + x. Optimizing the residual mapping is simpler than optimizing the original unreferenced mapping. In the extreme, it would be easier to drive the residual to zero than to fit an identity mapping with a stack of nonlinear layers if the identity mapping was optimal [43]. The skip connections execute identity mapping, which reduces training loss and their outputs are added to the outputs of stacked layers; hence, it does not increase computational complexity. This benefits in training very deep networks by avoiding superfluous connections. There are no parameters in the identity mapping as it simply adds the output from the previous layer to the next layer [42]. In-depth evaluation shows that ResNet outperformed other CNNs by obtaining a lowest top 5% error rate at 3.57% for classification problem. Hence, it can be said that the ResNets have gained popularity for image identification and classification problems due to their ability to overcome exploding/vanishing gradient problems while adding layers to an already deep neural network [41].Fig. 4 Residual learning [42] Because of Residual Networks’ superior performance in image classification, we used two ResNet architectures, ResNet18 and ResNet50, to extract deep features from CXR images. ResNet18 has 18 learnable layers, including 17 convolutional layers, a fully connected layer, and an additional softmax layer for classification. The kernel size for the convolutional layers is 3 × 3. The residual shortcut connections of ResNet18 skip two layers. The architecture of ResNet-18 is shown in Fig. 5. The ResNet50 architecture, on the other hand, contains 50 learnable layers, 48 of which are convolutional layers, a single average pooling and max-pooling layer. Unlike ResNet18, the ResNet50’s shortcut connections skip three layers rather than two. Figure 6 depicts the ResNet50 architecture. It may be noted that both ResNets have two sorts of shortcut connections. When the input and output dimensions are the same, the connections marked by straight arrows are adopted. When these dimensions expand, the other connections shown by the dotted lines are used. Figure 7 explains the detailed architecture specifications of ResNet18 and ResNet50 including details of learnable layers, their dimensions, strides and output sizes. Classification Fig. 5 ResNet18 architecture [44] Fig. 6 ResNet50 architecture [45] Fig. 7 Residual network architecture specifications [46] The deep feature vectors obtained from the Residual Networks are classified using SVM classifier with different kernel functions. SVM is a ML algorithm widely used in classification and regression tasks because of its stable performance. It belongs to the generalized linear classifier family that maximizes the margin between the hyperplane and the dataset to boost accuracy and avoid data overfitting. It searches for the dividing hyperplane that separates the two classes. This study employed SVM with four kernel functions, namely Linear, Cubic, Gaussian and Quadratic for classification. SVM techniques employ a set of mathematical functions known as kernels that take data as input and transform it into the desired form. The kernel functions return the scalar product of two points in highly appropriate feature space [47, 48]. Results and discussion Dataset The dataset used in this study is acquired from Kaggle. It consists of 15,161 CXR images distributed in four classes, i.e., COVID-19, Viral Pneumonia, Lung Opacity, and Normal. We only used Normal and COVID-19 CXR images in this study. The images in the database have a resolution of 299 × 299 pixels and are saved in the Portable Network Graphics (PNG) format [49]. The summary of the dataset is presented in Table 2. Figure 8 shows CXR samples from the dataset. We randomly divided the image dataset in the ratio of 70:30 for training and validation, respectively.Table 2 Dataset summary Class No of images Normal 10,192 COVID-19 3616 Fig. 8 Various COVID-19 and normal CXR scans Quantification is the process of determining the relative frequency (or prevalence) of the classes of interest in a dataset. It is commonly used in data analytics and classification applications to identify whether a population segment belongs to a specific class. Some popular online quantification methods include NEMSIS, CAN, SCAN [50]. In this paper, we performed exploratory data analysis to understand the class distribution in the dataset using pie chart as shown in Fig. 9. It may be noted that we only used Normal and COVID-19 CXR images in this study. Healthy image samples compose 78% of the entire dataset. The chart shows class imbalance issues in the dataset. Our major goal is to identify COVID-19 samples to facilitate early diagnosis of the disease, hence leading to a quicker treatment. The chart clearly indicates that COVID-19 samples account for only 24% of the overall dataset, hence in such circumstances, Recall, F1-Score and Precision are more appropriate metrics than accuracy. Evaluation parameters Fig. 9 Data distribution ratio Model performance evaluation is crucial for developing automated systems. Given that the major goal of such a model is to predict unforeseen data accurately, thus, evaluating the training and validation test sets indicate the model's generalization capabilities. In this aspect, a confusion matrix is an important metric for aiding in evaluating a classification model. The confusion matrix is a simple cross-tabulation of actual and predicted class values for all observations in each category. Various classification measures, such as precision, recall, and f1-score based on the confusion matrix, are used as a benchmark for evaluating the proposed model’s capability. Classification accuracy is commonly used metric since it summarizes the model’s performance. The f1-score, on the other hand, integrates precision and recall into a single metric that incorporates both properties. Precision appears in Eq. (1), Recall in Eq. (2), Accuracy in Eq. (3) and F1-Score in Eq. (4). 1 Precision=TP/TP+FP 2 Recall=TP/TP+FN 3 Accuracy=TP+TN/TP+TN+FP+FN 4 F1-Score=TP/TP+12FP+FN where TP = True Positives, TN = True Negative, FP = False Positive, FN = False Negative. Proposed method results This section presents the results of the experiments carried out in this study. All the experiments are performed on a machine with Intel Core i5 processor and 8 GB RAM using Deep Learning and Machine Learning Toolboxes of Matlab R2021a. We used two CNN architectures (ResNet18 & ResNet50) to extract deep features supplied to SVM with different kernels (Linear, Cubic, Quadratic and Gaussian) for classification. Table 3 shows the obtained accuracy of feature vectors obtained via different kernel functions. The SVM kernel types are shown in the rows and the Residual Network models in the columns. The average accuracy scores are shown in the last row and column of the table. ResNet18 and SVM classifier with Quadratic kernel achieved the maximum accuracy of 96.4%. The Cubic and Quadratic kernel functions, on the other hand, achieved 97.3% on classifying feature vectors obtained from ResNet50. According to the table, ResNet50 model provided the highest average accuracy of 96.6%, while the ResNet18 model produced an average accuracy of 95.7%. When the results are reviewed in terms of the kernel functions, it is clear that the Cubic kernel function produced the best overall accuracy of 96.9%, whereas the Quadratic-kernel-based SVM classifier produced the second-best average accuracy score of 96.8%. Other kernel functions such as Linear and Gaussian achieved an accuracy of 94.2% and 96.7%, respectively. The proposed method is capable of detecting COVID-19 from CXR images within 3–5 s.Table 3 Accuracy scores of residual networks and SVM classifiers on COVID-19 classification Accuracy % SVM Kernel/CNN ResNet18 ResNet50 Average Linear 93.4 94.9 94.2 Cubic 96.6 97.3 96.9 Quadratic 96.4 97.3 96.8 Gaussian 96.3 97.0 96.7 Average 95.7 96.6 96.6 Figure 10 shows the confusion matrix of the highest performing kernel on the feature vector obtained from ResNet18. The confusion matrices are used to summarize the prediction outcomes of a classification problem. It demonstrates the ways the classification model becomes confused when making new predictions. The X-axis of the matrix indicates the target class, while the Y-axis represents the output class. SVM correctly classified 987 COVID-19 samples and 3014 normal samples according to the confusion matrix. However, it misclassified 43 COVID-19 samples and 98 normal samples. Thus, obtaining the classification accuracy of 96.6%.Fig. 10 Confusion matrix of ResNet18 and cubic SVM classifier Figure 11 presents the confusion matrix of best-performing kernels (Cubic and Quadratic) on ResNet50’s feature vector. The Cubic kernel correctly identified 1015 COVID-19 samples and 3017 normal sample but misclassified 40 COVID-19 samples and 70 normal samples. Thus, obtaining classification accuracy of 97.34%, whereas Quadratic kernel accurately classified 1017 COVID-19 and 3015 normal samples and misdiagnosed 42 COVID-19 and 68 normal samples, achieving an accuracy of 97.34%.Fig. 11 Confusion matrix of ResNet50 and a Cubic SVM b Quadratic SVM The performance of our proposed technique is also assessed using precision, recall, and f1-score, as shown in Fig. 12. The ResNet50-based classification model outperformed the ResNet18-based classification model in terms of overall performance due to more extraction of more complicated features given the depth of the network design. Cross dataset validation Fig. 12 Precision, recall and F1-score values obtained for ResNet18 and ResNet50 The provided method’s robustness is evaluated using a cross-dataset validation scenario to demonstrate its applicability in real-world settings. The performance of the proposed technique is assessed using CXR images from the Kaggle dataset [51]. Our work accurately identified 72/77 COVID-19 samples and 77/77 Normal CXR samples during the cross-dataset validation process. As a result of the cross-database validation, we can infer that our method can be used in real-time to detect the presence of COVID-19 in CXR images. Comparison with state-of-the-art techniques Table 4 compares the performance of our proposed COVID-19 detection method with existing techniques in terms of accuracy. Xu et al. [23] deployed ResNet18 and achieved 90% accuracy. Hemdan et al. [24] fine-tuned VGG16 architecture to detect the presence of COVID-19 in CXR images and achieved 90% accuracy. However, in order to generalize better on real-world circumstances, the technique requires training on a larger dataset. The authors in Wang et al. [20] developed a 19 layered CNN architecture based on the idea of Residual Networks to discriminate between normal, bacterial infection, non-COVID-19 viral infection and COVID-19 viral infection. The system obtained the highest classification accuracy of 93.3%. However, the technique is not robust to images with intensity variations. In another study, Ismael et al. [28] developed an end-to-end CNN model containing 21 layers to classify COVID-19 that achieved 91.5% classification accuracy. The technique, however, obtained low overall accuracy. Hence, it must be thoroughly evaluated before being used for real-time COVID-19 detection. On the other hand, our proposed method achieved 96.4% accuracy using ResNet18 and 97.3% using ResNet50. The provided results demonstrate the suggested method’s efficiency and robustness when compared to existing methodologies.Table 4 Result comparison with state-of-the-art techniques Reference DL technique ACC % Xu et al. [23] ResNet18 86.7 Hemdan et al. [24] VGG16 90.0 Wang et al. [20] CNN 93.3 Ismael et al. [28] CNN 91.5 Proposed ResNet18 96.4 ResNet50 97.3 Conclusion This paper introduces a Residual Network-based automatic COVID-19 identification and classification system. The images in the dataset are first preprocessed before being fed into ResNet18 and ResNet50 to extract deep CNN features. The SVM classifier with different kernel functions is then used to classify the deep features vectors. Various assessment measures, such as Accuracy, Precision, F1-Score and Recall, are used to evaluate the suggested method's performance. With 96.4% accuracy and 97.3% accuracy, SVM classified the deep feature vector derived from ResNet18 and ResNet50, respectively. Moreover, we also evaluated the performance of our proposed method on a cross-dataset scenario. The proposed technique is robust and can detect COVID-19 in real-time. However, the dataset used in this study is imbalanced. Generally, a balanced data set with an equal class distribution produces a higher prediction accuracy. It may be noted that a balanced dataset makes it easier for the classification system to learn. However, the class imbalance was obvious because the images in the dataset were gathered from publically available datasets. Hence, additional COVID-19 CXR images will be collected in the future, and other CNN models will be explored for COVID-19 detection. Moreover, other lung illnesses will also be studied in the future, with the COVID-19 disease displaying different stages of mutation and imagistic patterns. Data availability All data generated or analyzed during this study are included in this published article. Declarations Conflicts of interest The authors declare that they have no conflicts of interest to report regarding the present study. Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References 1. Nayak SR Nayak DR Sinha U Arora V Pachori RB Application of deep learning techniques for detection of COVID-19 cases using chest X-ray images: a comprehensive study Biomed Signal Process Control 2021 64 102365 10.1016/j.bspc.2020.102365 33230398 2. Ayoobi N Time series forecasting of new cases and new deaths rate for COVID-19 using deep learning methods Results Phys 2021 27 104495 10.1016/j.rinp.2021.104495 34221854 3. Worldometer. (2021, 2022). CoronaVirus. Available: https://www.worldometers.info/coronavirus/ 4. Ghinai I et al (2020) First known person-to-person transmission of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) in the USA 5. Petrosillo N, Viceconte G, Ergonul O, Ippolito G, Petersen EJCM, and Infection (2020) COVID-19, SARS and MERS: are they closely related? 6. (2020) Symptoms of coronavirus. Available: https://www.cdc.gov/coronavirus/2019-ncov/symptoms-testing/symptoms.html 7. Badawi A Elgazzar K Detecting coronavirus from chest X-rays using transfer learning COVID 2021 1 1 403 415 10.3390/covid1010034 8. Xu T et al. (2020) Clinical features and dynamics of viral load in imported and non-imported patients with COVID-19 9. (2020) Available: https ://www.who.int/health-topics/coronavirus 10. Rousan LA Elobeid E Karrar M Khader Y Chest x-ray findings and temporal lung changes in patients with COVID-19 pneumonia BMC Pulm Med 2020 20 1 1 9 10.1186/s12890-020-01286-5 31900129 11. Cascella M, Rajnik M, Cuomo A, Dulebohn SC, Di Napoli R (2020) Features, evaluation and treatment coronavirus (COVID-19). Statpearls [internet] 12. Alafif T Tehame AM Bajaba S Barnawi A Zia S Machine and deep learning towards COVID-19 diagnosis and treatment: survey, challenges, and future directions Int J Environ Res Public Health 2021 18 3 1117 10.3390/ijerph18031117 33513984 13. Maguolo G Nanni L A critic evaluation of methods for covid-19 automatic detection from x-ray images Inf Fusion 2021 76 1 7 10.1016/j.inffus.2021.04.008 33967656 14. Ai T et al. (2020) Correlation of chest CT and RT-PCR testing in coronavirus disease 2019 (COVID-19) in China: a report of 1014 cases. 200642 15. Hall LO, Paul R, Goldgof DB, Goldgof GM (2020) Finding covid-19 from chest x-rays using deep learning on a small dataset. arXiv preprint arXiv:2004.02060 16. Jin Y-H A rapid advice guideline for the diagnosis and treatment of 2019 novel coronavirus (2019-nCoV) infected pneumonia (standard version) Mil Med Res 2020 7 1 4 32029004 17. Hussain E Hasan M Rahman MA Lee I Tamanna T Parvez MZ CoroDet: a deep learning based classification for COVID-19 detection using chest X-ray images Chaos Solitons Fractals 2021 142 110495 10.1016/j.chaos.2020.110495 33250589 18. Ibrahim AU, Ozsoz M, Serte S, Al-Turjman F, Yakoi PS (2021) Pneumonia classification using deep learning from chest X-ray images during COVID-19. Cogn Comput 1–13 19. Zu ZY, et al (2020) Coronavirus disease 2019 (COVID-19): a perspective from China. 200490 20. Wang L Lin ZQ Wong A Covid-net: a tailored deep convolutional neural network design for detection of covid-19 cases from chest x-ray images Sci Rep 2020 10 1 1 12 31913322 21. Ahammed K, Satu MS, Abedin MZ, Rahaman MA, Islam SMS (2020) Early detection of coronavirus cases using chest x-ray images employing machine learning and deep learning approaches. medRxiv 22. Kibriya H, Rafique R, Ahmad W, Adnan S (2021) Tomato leaf disease detection using convolution neural network. In: 2021 international bhurban conference on applied sciences and technologies (IBCAST). IEEE, pp 346–351 23. Xu X A deep learning system to screen novel coronavirus disease 2019 pneumonia Engineering 2020 6 10 1122 1129 10.1016/j.eng.2020.04.010 32837749 24. Hemdan EE-D, Shouman MA, Karar ME (2020) Covidx-net: a framework of deep learning classifiers to diagnose covid-19 in x-ray images. arXiv preprint arXiv:2003.11055 25. Singh D Kumar V Kaur M Classification of COVID-19 patients from chest CT images using multi-objective differential evolution–based convolutional neural networks Eur J Clin Microbiol Infect Dis 2020 39 7 1379 1389 10.1007/s10096-020-03901-z 32337662 26. Loey M Smarandache F Khalifa NEM Within the lack of COVID-19 benchmark dataset: a novel gan with deep transfer learning for corona-virus detection in chest x-ray images Symmetry 2020 12 4 1 19 10.3390/sym12040651 27. Hernandez D, Pereira R, Georgevia P (2020) COVID-19 detection through x-ray chest images. In: 2020 international conference automatics and informatics (ICAI). IEEE, pp 1–5 28. Ismael AM Şengür A Deep learning approaches for COVID-19 detection based on chest x-ray images Expert Syst Appl 2021 164 114054 10.1016/j.eswa.2020.114054 33013005 29. Narin A, Kaya C, Pamuk ZJAPA (2020) Automatic detection of coronavirus disease (covid-19) using x-ray images and deep convolutional neural networks. 30. Wang S, et al. (2021) A deep learning algorithm using CT images to screen for Corona Virus Disease (COVID-19). Eur Radiol 1–9 31. Pathak Y, Shukla PK, Tiwari A, Stalin S, Singh S (2020) Deep transfer learning based classification model for COVID-19 disease. Irbm 32. Zhao J, Zhang Y, He X, Xie P (2020) Covid-ct-dataset: a ct scan dataset about covid-19. arXiv preprint arXiv:2003.13865 33. Bai X Predicting COVID-19 malignant progression with AI techniques SSRN J 2020 10.2139/ssrn.3557984 34. Bora DJ (2017) Importance of image enhancement techniques in color image segmentation: a comprehensive and comparative study. arXiv preprint arXiv:1708.05081 35. Hubel DH Wiesel TN Receptive fields, binocular interaction and functional architecture in the cat's visual cortex J Physiol 1962 160 1 106 154 10.1113/jphysiol.1962.sp006837 14449617 36. Marr D (1977) Representing visual information 37. Brownlee J (2019) A gentle introduction to the rectified linear unit (ReLU). Available: https://machinelearningmastery.com/rectified-linear-activation-function-for-deep-learning-neural-networks/ 38. Brownlee J (2019) A gentle introduction to pooling layers for convolutional neural networks. Available: https://machinelearningmastery.com/pooling-layers-for-convolutional-neural-networks/ 39. Hassanien AE Gaber T Handbook of research on machine learning innovations and trends 2017 Pennsylvania IGI global 40. Ouchicha C Ammor O Meknassi M CVDNet: a novel deep learning architecture for detection of coronavirus (Covid-19) from chest x-ray images Chaos Solitons Fractals 2020 140 110245 10.1016/j.chaos.2020.110245 32921934 41. (2020) Introduction to residual networks. Available: https://www.geeksforgeeks.org/introduction-to-residual-networks/ 42. He K, Zhang X, Ren S, Sun J (2016) Deep residual learning for image recognition. In: Proceedings of the IEEE conference on computer vision and pattern recognition. pp 770–778 43. Dietz M (2017) Understand deep residual networks—a simple, modular learning framework that has redefined state-of-the-art. Available: https://medium.com/@waya.ai/deep-residual-learning-9610bb62c355 44. Ramzan F A deep learning approach for automated diagnosis and multi-class classification of Alzheimer’s disease stages using resting-state fMRI and residual neural networks J Med Syst 2020 44 2 1 16 10.1007/s10916-019-1475-2 45. Abd ElGhany S Ibraheem MR Alruwaili M Elmogy M Diagnosis of various skin cancer lesions based on fine-tuned ResNet50 deep network CMC-Comput Mater Contin 2021 68 1 117 135 10.32604/cmc.2021.016102 46. Wu H Xin M Fang W Hu H-M Hu Z Multi-level feature network with multi-loss for person re-identification IEEE Access 2019 7 91052 91062 10.1109/ACCESS.2019.2927052 47. Suthaharan S Support vector machine Machine learning models and algorithms for big data classification 2016 New York Springer 207 235 48. Xue H Yang Q Chen S SVM: support vector machines The top ten algorithms in data mining 2009 Boca Raton Chapman and Hall/CRC 51 74 49. Kaggle (ed) COVID-19 radiography database 50. Kar P, Li S, Narasimhan H, Chawla S, Sebastiani F (2016) Online optimization methods for the quantification problem. In: Proceedings of the 22nd ACM SIGKDD international conference on knowledge discovery and data mining, pp 1625–1634. 51. Patel P, Kaggle (eds) (2021) Chest X-ray (Covid-19 & Pneumonia)

0

PMC9754308

NO-CC CODE

2022-12-16 23:26:35

no

Neural Comput Appl. 2022 Dec 15;:1-12

utf-8

Neural Comput Appl

10.1007/s00521-022-08127-y

oa_other

==== Front Neural Comput Appl Neural Comput Appl Neural Computing & Applications 0941-0643 1433-3058 Springer London London 8127 10.1007/s00521-022-08127-y Original Article A residual network-based framework for COVID-19 detection from CXR images Kibriya Hareem hareem.kibriya@students.uettaxila.edu.pk 1 http://orcid.org/0000-0002-3143-689X Amin Rashid rashid.amin@uettaxila.edu.pk 12 1 grid.442854.b Department of Computer Sciences, University of Engineering and Technology, Taxila, Pakistan 2 Department of Computer Science, University of Chakwal, Chakwal, 48800, Pakistan 15 12 2022 112 17 11 2021 28 11 2022 © The Author(s), under exclusive licence to Springer-Verlag London Ltd., part of Springer Nature 2022, Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. In late 2019, a new Coronavirus disease (COVID-19) appeared in Wuhan, Hubei Province, China. The virus began to spread throughout many countries, affecting a large population. Polymerase chain reaction is currently being utilized to diagnose COVID-19 in suspected patients; however, its sensitivity is quite low. The researchers also developed automated approaches for reliably and timely identifying COVID-19 from X-ray images. However, traditional machine learning-based image classification algorithms necessitate manual image segmentation and feature extraction, which is a time-consuming task. Due to promising results and robust performance, Convolutional Neural Network (CNN)-based techniques are being used widely to classify COVID-19 from Chest X-rays (CXR). This study explores CNN-based COVID-19 classification methods. A series of experiments aimed at COVID-19 detection and classification validates the viability of our proposed framework. Initially, the dataset is preprocessed and then fed into two Residual Network (ResNet) architectures for deep feature extraction, such as ResNet18 and ResNet50, whereas support vector machines with its multiple kernels, including Quadratic, Linear, Gaussian and Cubic, are used to classify these features. The experimental results suggest that the proposed framework efficiently detects COVID-19 from CXR images. The proposed framework obtained the best accuracy of 97.3% using ResNet50. Keywords COVID-19 Machine learning CXR images SVM ResNet50 ==== Body pmcIntroduction Coronaviruses have been responsible for two large-scale outbreaks in the last two decades: SARS and MERS. COVID-19, also known as SARS-CoV-2, is a novel human coronavirus that arose in Wuhan in Hubei Province, China, in late 2019 [1]. This virus has also been reported in almost all continents, including Europe, Australia, Asia, and America [2]. Until the 16th of February, 2022, 415 million confirmed cases had been reported worldwide, with a total of 5.84 million deaths [3]. A study shows the zoonotic origin of COVID-19, and the fast expansion suggests continuous person-to-person transmission [4]. COVID-19 is a respiratory disease characterized by symptoms ranging from a moderate infection in upper respiratory system to acute respiratory distress syndrome (ARDS) [5]. COVID-19 symptoms might differ from person to person, but some are typical. These symptoms typically emerge between 3 and 14 days. The Center for Disease Control and Prevention (CDC) [6] lists major COVID-19 symptoms: dry cough, trouble breathing, fever, muscle discomfort, sore throat, loss of smell and taste. Some of the less common symptoms are nausea, vomiting, and diarrhea. This virus has been shown to be airborne for up to 3 h, 4 h on copper surfaces, and nearly 72 h on plastic and stainless materials. However, identifying the actual nature of the virus remains an unresolved topic in the medical research world [7]. Due to surge in cases, the World Health Organization (WHO) declared COVID-19 a pandemic and public health emergency of worldwide significance [8, 9]. Early and precise diagnosis of acute COVID-19 is necessary to reduce healthcare capacity and overall death rate [10]. Since the emergence of infections, the world governments have put plenty of countermeasures to mitigate the devastation being caused. Researchers and health organizations all over the world are working feverishly to develop early disease detection and treatment approaches [11]. Scientists from all around the world have developed various COVID-19 vaccines. Many of these vaccines have already been approved and are being used for immunization in numerous countries worldwide. These vaccines include Pfizer-BioNTech, Moderna, Oxford-AstraZeneca, Sinovac and Sputnik V. Despite widespread vaccination campaigns, coronavirus infections have risen, owing primarily to new variants. However, in many countries, mass vaccine adaptation remains a problem of public health logistics (e.g., manufacturing, storing, and distributing the vaccine and mass immunization) and leadership. The problem is ascribed to various factors, including vaccination resistance among citizens and vaccine nationalism [12, 13]. Further study is required to address outstanding questions such as: would these vaccines be able to manage the pandemic? What is the efficacy of these vaccines against developing various COVID-19 variants? What are the adverse effects of contemporary vaccines, studied, developed, and tested at a fast pace on various population statistics? [7] RT-PCR tests must validate the diagnosis of COVID-19 according to the recommendations by the Chinese government [14]. However, it has some limitations, such as long delays in obtaining the results and low sensitivity. The patients with high clinical suspicion of the virus test falsely negative on initial RT-PCR tests require multiple tests to validate the results [15]. Low test sensitivity may occur due to the following factors: suboptimal clinical sample procedures, fluctuations in viral load, and manufacturer test kit sensitivity. With communities experiencing an increase in patients, negative RT-PCR test is becoming very difficult to handle [16]. According to October 2020 recommendations by WHO, chest imaging evaluation is an effective method for detecting clinical signs of COVID-19 in patients because the infection is related to the pulmonary system [17]. The study presented by Ai et al. [14] compares the sensitivity of the Chest computed tomography (CT) and the RT-PCR tests. The Chest CT obtained a better sensitivity score than RT-PCR for detecting COVID-19 in suspected patients. Other imaging modalities include X-rays, MRIs, and needle biopsy of the lung. Compared to the CT Scans, Chest X-ray is the most commonly utilized to identify COVID-19 because CT imaging is expensive, and the high ionizing radiation levels may pose health hazards to pregnant women and children. Moreover, it takes longer to scan image and is scarcely available in low-income countries [18]. On the other hand, X-ray imaging is widely used in many medical and epidemiological applications due to its widespread availability [19, 20]. Due to ease of use by radiologists, lightweight operating speed and cheaper cost, the chest X-ray is a well-suited imaging modality for diagnosing pulmonary infections [21]. Over the past two years, various frameworks employing CNNs have been presented to detect COVID-19 from chest X-Ray images. CNN is a type of Deep Neural Networks (DNN) inspired by the biological working of the human brain and used to analyze visual imagery. CNNs are extremely efficient and require little or no preprocessing before classification. These architectures automatically extract deep features and perform classification, whereas typical Machine Learning (ML) algorithms require a manual segmentation and handcrafted feature extraction. This independence of CNNs from human interaction and prior information is a significant asset. Doctors can benefit from automated disease classification systems during the treatment procedure [22]. Several studies have been proposed to detect COVID-19 and other lung illnesses using Chest CT and X-rays; however, these techniques have limited classification accuracy and a large false-positive rate due to poor data quality and inefficient methods. Hence, there is a need for a robust approach that can distinguish between COVID-19 and normal images efficiently. As a result, we present a novel COVID-19 identification and classification approach based on well-known ResNet architectures for deep feature extraction and SVM for image classification. The contributions of the paper are as follows:The input CXR images are initially preprocessed to enhance the essential features helpful for classification. We improved the contrast of dataset images and resized them to fit the CNNs’ input layer size. The deep CNN features are extracted using final Fully Connected layers from Residual Network architectures such as ResNet18 and ResNet50. The SVM classifier with four kernel functions such as Quadratic, Linear, Gaussian, and Cubic has been used to classify the deep CNN features. The proposed method is compared to existing state-of-the-art COVID-19 classification and identification models to determine its robustness. The performance of the suggested method is evaluated using various measures such as recall, precision, f1-score, and accuracy. The proposed technique is also validated on a cross-dataset validation scenario. Related work Artificial intelligence and DL-based new methodologies have significantly impacted medical image analysis, particularly in disease detection. Since the emergence of COVID-19, researchers have been developing DL and ML based techniques for early diagnosis of COVID-19 using Chest CTs and X-rays. However, accurate diagnosis is still a challenge due to the lack of quality data and robust techniques. Xu et al. [23] provided an early prediction approach for identifying COVID-19 using ResNet18 that obtained 86.7% accuracy. Hemdan et al. [24] deployed a pre-trained CNN model, i.e., VGG16, to discriminate between COVID-19 and Normal images with 90% accuracy. Singh et al. [25] presented a novel multi-objective differential evolution-based CNN architecture for distinguishing COVID-19 from normal CT images. To detect COVID-19 from images, Loey et al. [26] used pre-trained CNNs such as GoogLeNet, AlexNet, and ResNet18 combined with Generative Adversarial Networks (GAN). Hernandez et al. [27] suggested a COVID-19 detection and classification system using fine-tuned CNN architectures, including ResNet, VGG, and DenseNet, and reported an overall accuracy of 90%. COVID-Net, a 19-layered CNN architecture based on the principle of Residual Networks, was created by Wang et al. [20] to distinguish between normal images, bacterial infection images, and non-COVID-19 viral, and COVID-19 viral infection images. COVID-Net provided the model with the highest classification accuracy of 93.3%. In another study, Ismael et al. [28] developed an end-to-end CNN model composed of 21 layers to classify COVID-19. The novel CNN architecture achieved 91.5% classification accuracy. Despite creating a modified CNN, the system's overall accuracy was low. In another study, Narin et al. [29] proposed a ResNet50, Inception-ResNet V2 and Inception-V3 based approach to classify COVID-19. However, in order to generalize the approach to real-world scenarios, it must be trained and evaluated on datasets with more image samples. Wang et al. [30] employed pre-trained CNN architectures to discriminate between COVID-19 and healthy images. The study reported 89.5% classification accuracy. Pathak et al. [31] employed various transfer learned residual networks architectures to classify COVID-19 infected images. They obtained the highest accuracy of 93% on publically available CT image datasets. However, the system is not robust in classifying high-intensity variations noisy images. Zhao et al. [32] applied transfer learned DenseNet architecture with 169 layers to identify the presence of COVID-19 from CT images. The system achieved 84.7% accuracy. However, the system is computationally exhausting. The authors in Ibrahim et al. [18] fine-tuned AlexNet to identify and classify COVID-19 from images, but they did not equate the performance of their proposed system to current approaches. On the other hand, Bai et al. [33] suggest a COVID-19 detection approach employing Multi-layered perceptron with LSTM. However, the system is resource-intensive due to the combination of LSTM and multi-layered perceptron. Table 1 provides a summary of the existing COVID-19 identification methods, as well as their limitations.Table 1 Summary of existing COVID-19 identification systems Reference DL model Limitations Xu et al. [23] ResNet18 The technique needs to be trained and evaluated on a larger dataset to generalize better on real scenarios Wang et al. [30] Inception The work is not robust to images with noise and intensity variation Hernandez et al. [27] ResNet, VGG, DenseNet The method needs performance improvement Hemdan et al. [24] VGG16 They used a very limited dataset with only 100 images. Hence, the technique needs to be validated Singh et al.[25] CNN The model needs to be trained and evaluated on a large database Zhao et al. [32] DenseNet The network uses a lot of memory thus is computationally expensive Wang et al. [20] CNN The technique may not be able to classify images with intensity variations Pathak et al. [31] Residual Networks The work is not robust to the classification of images with noise and contrasting intensity values Bai et al. [33] Multi-layered perceptron with LSTM Combining LSTM and multi-layered perceptron increases the parameters; thus the inference time will be high Ismael et al. [28] CNN, ResNet, VGG16, 19 The technique may fail to perform on images with intensity variation. Moreover, it needs a larger database for system training and evaluation Loey et al. [26] GoogLeNet, AlexNet and ResNet18, GAN They used a very small dataset containing only 306 samples for four classes, i.e., normal, pneumonia (viral and bacterial) and COVID-19 Ibrahim et al. [18] AlexNet The proposed method's performance is not compared to state-of-the-art techniques Narin et al. [29] Inception-V3, Inception-ResNet V2, ResNet50 The method may not perform well in real-time because it is trained and evaluated in a very small dataset Proposed methodology In this paper, we propose a DL based technique to detect the presence of COVID-19 in CXR images. The detailed block diagram of the proposed method is shown in Fig. 1. Initially, the dataset images are preprocessed before being fed into two Residual Network models for deep CNN feature extraction, such as ResNet18 or ResNet50. The deep feature vectors are classified using SVM with different kernel functions: Linear, Gaussian, Cubic, and Quadratic.Data preprocessing Fig. 1 Diagram illustrating the proposed framework This section discusses the preprocessing techniques applied to dataset images before feeding them to CNNs for feature extraction. Initially, the contrast of CXR images in dataset is enhanced by mapping the intensity values of an input image to new values after saturating the top and bottom 1% of the pixel values. Figure 2 depicts a CXR image before and after contrast enhancement. Contrast enhancement is a technique that alters the pixel intensity of an image to use as many bins as possible. It distinguishes between dark and bright areas in an image and eliminates any ambiguity that might otherwise exist between different image regions [34]. The images are then resized to match the input layer size of CNNs. Deep feature extraction using residual networks Fig. 2 Image before and after contrast enhancement CNNs are widely used in various research fields due to their superior performance over traditional ML-based techniques. The architecture of CNN is biologically inspired by the human brain area called the visual cortex that contains cells sensitive to visual perception. The use of CNNs in ML derives from a 1962 study by Hubel and Weisel, who discovered that some individual neuron cells in the brain that activate only in the presence of specific edges and shapes [35]. Usually, the CNN architectures have numerous layers, as shown in Fig. 3 (convolution, pooling and fully connected) stacked on top of each other. These models learn more complicated features with an increase in architecture’s depth. The convolution layer extract features from an image by computing the dot product between the image and the kernel. The kernel slides through the image region to compute the features, bypassing particular pixels called stride. As first described by Marr, the filter detects various low-level features such as colors, edges, curves, boundaries, virtual lines and high-level features such as local surfaces, orientations, and discontinuities [36]. The convolved portion of the image is called a receptive field. The features extracted during this process serve as input to deep layers in the architecture. This layer is followed by an activation function that is responsible for converting the node's summed weighted input into the node's activation for that input. Due to ease in training and better performance, ReLU has become a default activation function for neural networks. It is responsible for reducing all negative values to zero [37].Fig. 3 General architecture of the CNN One of the limitations of convolution layer’s feature map output is that it records the precise position details of the features in an input. Thus, very minor changes in the position of the feature in an image can result in a distinct feature map. This can occur when the input image is rotated, re-cropped, shifted, etc. Hence, downsampling is frequently used to create a reduced resolution version of an input signal that maintains structural parts of an image but lacks very fine or minor details that may be invaluable for the task. This downsampling process is performed using a Pooling layer that is usually placed after a Convolution layer. It decreases the spatial dimensions of an image following the convolution process. If we apply the Fully Connected Layer straight after the Convolution Layer, the system may become computationally expensive due to the high dimensionality of the image. As a result, pooling layers are utilized to scale the images and lower the dataset's spatial volume [38]. The output obtained from the convolution layers usually represents high-level features in the data. This output is generally flattened and connected to the output layer by adding a fully-connected layer to learn nonlinear combinations of these features. All neurons in a fully connected layer connect to all neurons in the previous layer. This layer combines all of the local features gathered by the preceding layers across the image to discover bigger patterns. The classifier then performs classification based on these features [39, 40]. Over the past few years, remarkable development has been made in Deep Learning. Following CNN architecture AlexNet, which won the ImageNet 2012 competition, each successive winning architecture employs more layers in a deep neural network to lower error rates. Deep Neural Networks are becoming more complex and deeper with time. It has been demonstrated that increasing the number of layers in these networks increases its performance and robustness for image-related tasks. Hence, the researchers increase the depth of the neural networks to extract important features from complex images. As a result, the initial layers may identify edges, and succeeding layers may detect complete objects at the end. However, this works for networks with fewer layers, but with network convergence, the accuracy of these models tends to be saturated. This decrease in the accuracy score is not due to overfitting but rather to a rise in error rate after increasing the depth of the network, as current solvers are unable to optimize the network efficiently. This runs counter to the popular belief that increasing layers in the network improves its performance. It may be noted that this problem does not appear due to network overfitting because, in such scenario, the dropout and regularization techniques can be used to solve this problem. But it is mostly present because of the well-known Vanishing/Exploding gradient problem. This problem causes the gradients of the loss function to approach zero or very large when more layers with specific activation functions are added to neural networks, thus making the network difficult to train and increasing train and test errors [41, 42]. In 2015, the researchers at Microsoft proposed Residual Network (ResNet). The introduction of ResNets aided in the training of very deep networks. ResNets use shortcut connections to skip one or more layers, as shown in Fig. 4. The layers fit a residual mapping rather than assuming each stack of layers immediately fits a desired underlying mapping. The original mapping is transformed into F(x) + x. Optimizing the residual mapping is simpler than optimizing the original unreferenced mapping. In the extreme, it would be easier to drive the residual to zero than to fit an identity mapping with a stack of nonlinear layers if the identity mapping was optimal [43]. The skip connections execute identity mapping, which reduces training loss and their outputs are added to the outputs of stacked layers; hence, it does not increase computational complexity. This benefits in training very deep networks by avoiding superfluous connections. There are no parameters in the identity mapping as it simply adds the output from the previous layer to the next layer [42]. In-depth evaluation shows that ResNet outperformed other CNNs by obtaining a lowest top 5% error rate at 3.57% for classification problem. Hence, it can be said that the ResNets have gained popularity for image identification and classification problems due to their ability to overcome exploding/vanishing gradient problems while adding layers to an already deep neural network [41].Fig. 4 Residual learning [42] Because of Residual Networks’ superior performance in image classification, we used two ResNet architectures, ResNet18 and ResNet50, to extract deep features from CXR images. ResNet18 has 18 learnable layers, including 17 convolutional layers, a fully connected layer, and an additional softmax layer for classification. The kernel size for the convolutional layers is 3 × 3. The residual shortcut connections of ResNet18 skip two layers. The architecture of ResNet-18 is shown in Fig. 5. The ResNet50 architecture, on the other hand, contains 50 learnable layers, 48 of which are convolutional layers, a single average pooling and max-pooling layer. Unlike ResNet18, the ResNet50’s shortcut connections skip three layers rather than two. Figure 6 depicts the ResNet50 architecture. It may be noted that both ResNets have two sorts of shortcut connections. When the input and output dimensions are the same, the connections marked by straight arrows are adopted. When these dimensions expand, the other connections shown by the dotted lines are used. Figure 7 explains the detailed architecture specifications of ResNet18 and ResNet50 including details of learnable layers, their dimensions, strides and output sizes. Classification Fig. 5 ResNet18 architecture [44] Fig. 6 ResNet50 architecture [45] Fig. 7 Residual network architecture specifications [46] The deep feature vectors obtained from the Residual Networks are classified using SVM classifier with different kernel functions. SVM is a ML algorithm widely used in classification and regression tasks because of its stable performance. It belongs to the generalized linear classifier family that maximizes the margin between the hyperplane and the dataset to boost accuracy and avoid data overfitting. It searches for the dividing hyperplane that separates the two classes. This study employed SVM with four kernel functions, namely Linear, Cubic, Gaussian and Quadratic for classification. SVM techniques employ a set of mathematical functions known as kernels that take data as input and transform it into the desired form. The kernel functions return the scalar product of two points in highly appropriate feature space [47, 48]. Results and discussion Dataset The dataset used in this study is acquired from Kaggle. It consists of 15,161 CXR images distributed in four classes, i.e., COVID-19, Viral Pneumonia, Lung Opacity, and Normal. We only used Normal and COVID-19 CXR images in this study. The images in the database have a resolution of 299 × 299 pixels and are saved in the Portable Network Graphics (PNG) format [49]. The summary of the dataset is presented in Table 2. Figure 8 shows CXR samples from the dataset. We randomly divided the image dataset in the ratio of 70:30 for training and validation, respectively.Table 2 Dataset summary Class No of images Normal 10,192 COVID-19 3616 Fig. 8 Various COVID-19 and normal CXR scans Quantification is the process of determining the relative frequency (or prevalence) of the classes of interest in a dataset. It is commonly used in data analytics and classification applications to identify whether a population segment belongs to a specific class. Some popular online quantification methods include NEMSIS, CAN, SCAN [50]. In this paper, we performed exploratory data analysis to understand the class distribution in the dataset using pie chart as shown in Fig. 9. It may be noted that we only used Normal and COVID-19 CXR images in this study. Healthy image samples compose 78% of the entire dataset. The chart shows class imbalance issues in the dataset. Our major goal is to identify COVID-19 samples to facilitate early diagnosis of the disease, hence leading to a quicker treatment. The chart clearly indicates that COVID-19 samples account for only 24% of the overall dataset, hence in such circumstances, Recall, F1-Score and Precision are more appropriate metrics than accuracy. Evaluation parameters Fig. 9 Data distribution ratio Model performance evaluation is crucial for developing automated systems. Given that the major goal of such a model is to predict unforeseen data accurately, thus, evaluating the training and validation test sets indicate the model's generalization capabilities. In this aspect, a confusion matrix is an important metric for aiding in evaluating a classification model. The confusion matrix is a simple cross-tabulation of actual and predicted class values for all observations in each category. Various classification measures, such as precision, recall, and f1-score based on the confusion matrix, are used as a benchmark for evaluating the proposed model’s capability. Classification accuracy is commonly used metric since it summarizes the model’s performance. The f1-score, on the other hand, integrates precision and recall into a single metric that incorporates both properties. Precision appears in Eq. (1), Recall in Eq. (2), Accuracy in Eq. (3) and F1-Score in Eq. (4). 1 Precision=TP/TP+FP 2 Recall=TP/TP+FN 3 Accuracy=TP+TN/TP+TN+FP+FN 4 F1-Score=TP/TP+12FP+FN where TP = True Positives, TN = True Negative, FP = False Positive, FN = False Negative. Proposed method results This section presents the results of the experiments carried out in this study. All the experiments are performed on a machine with Intel Core i5 processor and 8 GB RAM using Deep Learning and Machine Learning Toolboxes of Matlab R2021a. We used two CNN architectures (ResNet18 & ResNet50) to extract deep features supplied to SVM with different kernels (Linear, Cubic, Quadratic and Gaussian) for classification. Table 3 shows the obtained accuracy of feature vectors obtained via different kernel functions. The SVM kernel types are shown in the rows and the Residual Network models in the columns. The average accuracy scores are shown in the last row and column of the table. ResNet18 and SVM classifier with Quadratic kernel achieved the maximum accuracy of 96.4%. The Cubic and Quadratic kernel functions, on the other hand, achieved 97.3% on classifying feature vectors obtained from ResNet50. According to the table, ResNet50 model provided the highest average accuracy of 96.6%, while the ResNet18 model produced an average accuracy of 95.7%. When the results are reviewed in terms of the kernel functions, it is clear that the Cubic kernel function produced the best overall accuracy of 96.9%, whereas the Quadratic-kernel-based SVM classifier produced the second-best average accuracy score of 96.8%. Other kernel functions such as Linear and Gaussian achieved an accuracy of 94.2% and 96.7%, respectively. The proposed method is capable of detecting COVID-19 from CXR images within 3–5 s.Table 3 Accuracy scores of residual networks and SVM classifiers on COVID-19 classification Accuracy % SVM Kernel/CNN ResNet18 ResNet50 Average Linear 93.4 94.9 94.2 Cubic 96.6 97.3 96.9 Quadratic 96.4 97.3 96.8 Gaussian 96.3 97.0 96.7 Average 95.7 96.6 96.6 Figure 10 shows the confusion matrix of the highest performing kernel on the feature vector obtained from ResNet18. The confusion matrices are used to summarize the prediction outcomes of a classification problem. It demonstrates the ways the classification model becomes confused when making new predictions. The X-axis of the matrix indicates the target class, while the Y-axis represents the output class. SVM correctly classified 987 COVID-19 samples and 3014 normal samples according to the confusion matrix. However, it misclassified 43 COVID-19 samples and 98 normal samples. Thus, obtaining the classification accuracy of 96.6%.Fig. 10 Confusion matrix of ResNet18 and cubic SVM classifier Figure 11 presents the confusion matrix of best-performing kernels (Cubic and Quadratic) on ResNet50’s feature vector. The Cubic kernel correctly identified 1015 COVID-19 samples and 3017 normal sample but misclassified 40 COVID-19 samples and 70 normal samples. Thus, obtaining classification accuracy of 97.34%, whereas Quadratic kernel accurately classified 1017 COVID-19 and 3015 normal samples and misdiagnosed 42 COVID-19 and 68 normal samples, achieving an accuracy of 97.34%.Fig. 11 Confusion matrix of ResNet50 and a Cubic SVM b Quadratic SVM The performance of our proposed technique is also assessed using precision, recall, and f1-score, as shown in Fig. 12. The ResNet50-based classification model outperformed the ResNet18-based classification model in terms of overall performance due to more extraction of more complicated features given the depth of the network design. Cross dataset validation Fig. 12 Precision, recall and F1-score values obtained for ResNet18 and ResNet50 The provided method’s robustness is evaluated using a cross-dataset validation scenario to demonstrate its applicability in real-world settings. The performance of the proposed technique is assessed using CXR images from the Kaggle dataset [51]. Our work accurately identified 72/77 COVID-19 samples and 77/77 Normal CXR samples during the cross-dataset validation process. As a result of the cross-database validation, we can infer that our method can be used in real-time to detect the presence of COVID-19 in CXR images. Comparison with state-of-the-art techniques Table 4 compares the performance of our proposed COVID-19 detection method with existing techniques in terms of accuracy. Xu et al. [23] deployed ResNet18 and achieved 90% accuracy. Hemdan et al. [24] fine-tuned VGG16 architecture to detect the presence of COVID-19 in CXR images and achieved 90% accuracy. However, in order to generalize better on real-world circumstances, the technique requires training on a larger dataset. The authors in Wang et al. [20] developed a 19 layered CNN architecture based on the idea of Residual Networks to discriminate between normal, bacterial infection, non-COVID-19 viral infection and COVID-19 viral infection. The system obtained the highest classification accuracy of 93.3%. However, the technique is not robust to images with intensity variations. In another study, Ismael et al. [28] developed an end-to-end CNN model containing 21 layers to classify COVID-19 that achieved 91.5% classification accuracy. The technique, however, obtained low overall accuracy. Hence, it must be thoroughly evaluated before being used for real-time COVID-19 detection. On the other hand, our proposed method achieved 96.4% accuracy using ResNet18 and 97.3% using ResNet50. The provided results demonstrate the suggested method’s efficiency and robustness when compared to existing methodologies.Table 4 Result comparison with state-of-the-art techniques Reference DL technique ACC % Xu et al. [23] ResNet18 86.7 Hemdan et al. [24] VGG16 90.0 Wang et al. [20] CNN 93.3 Ismael et al. [28] CNN 91.5 Proposed ResNet18 96.4 ResNet50 97.3 Conclusion This paper introduces a Residual Network-based automatic COVID-19 identification and classification system. The images in the dataset are first preprocessed before being fed into ResNet18 and ResNet50 to extract deep CNN features. The SVM classifier with different kernel functions is then used to classify the deep features vectors. Various assessment measures, such as Accuracy, Precision, F1-Score and Recall, are used to evaluate the suggested method's performance. With 96.4% accuracy and 97.3% accuracy, SVM classified the deep feature vector derived from ResNet18 and ResNet50, respectively. Moreover, we also evaluated the performance of our proposed method on a cross-dataset scenario. The proposed technique is robust and can detect COVID-19 in real-time. However, the dataset used in this study is imbalanced. Generally, a balanced data set with an equal class distribution produces a higher prediction accuracy. It may be noted that a balanced dataset makes it easier for the classification system to learn. However, the class imbalance was obvious because the images in the dataset were gathered from publically available datasets. Hence, additional COVID-19 CXR images will be collected in the future, and other CNN models will be explored for COVID-19 detection. Moreover, other lung illnesses will also be studied in the future, with the COVID-19 disease displaying different stages of mutation and imagistic patterns. Data availability All data generated or analyzed during this study are included in this published article. Declarations Conflicts of interest The authors declare that they have no conflicts of interest to report regarding the present study. Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References 1. Nayak SR Nayak DR Sinha U Arora V Pachori RB Application of deep learning techniques for detection of COVID-19 cases using chest X-ray images: a comprehensive study Biomed Signal Process Control 2021 64 102365 10.1016/j.bspc.2020.102365 33230398 2. Ayoobi N Time series forecasting of new cases and new deaths rate for COVID-19 using deep learning methods Results Phys 2021 27 104495 10.1016/j.rinp.2021.104495 34221854 3. Worldometer. (2021, 2022). CoronaVirus. Available: https://www.worldometers.info/coronavirus/ 4. Ghinai I et al (2020) First known person-to-person transmission of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) in the USA 5. Petrosillo N, Viceconte G, Ergonul O, Ippolito G, Petersen EJCM, and Infection (2020) COVID-19, SARS and MERS: are they closely related? 6. (2020) Symptoms of coronavirus. Available: https://www.cdc.gov/coronavirus/2019-ncov/symptoms-testing/symptoms.html 7. Badawi A Elgazzar K Detecting coronavirus from chest X-rays using transfer learning COVID 2021 1 1 403 415 10.3390/covid1010034 8. Xu T et al. (2020) Clinical features and dynamics of viral load in imported and non-imported patients with COVID-19 9. (2020) Available: https ://www.who.int/health-topics/coronavirus 10. Rousan LA Elobeid E Karrar M Khader Y Chest x-ray findings and temporal lung changes in patients with COVID-19 pneumonia BMC Pulm Med 2020 20 1 1 9 10.1186/s12890-020-01286-5 31900129 11. Cascella M, Rajnik M, Cuomo A, Dulebohn SC, Di Napoli R (2020) Features, evaluation and treatment coronavirus (COVID-19). Statpearls [internet] 12. Alafif T Tehame AM Bajaba S Barnawi A Zia S Machine and deep learning towards COVID-19 diagnosis and treatment: survey, challenges, and future directions Int J Environ Res Public Health 2021 18 3 1117 10.3390/ijerph18031117 33513984 13. Maguolo G Nanni L A critic evaluation of methods for covid-19 automatic detection from x-ray images Inf Fusion 2021 76 1 7 10.1016/j.inffus.2021.04.008 33967656 14. Ai T et al. (2020) Correlation of chest CT and RT-PCR testing in coronavirus disease 2019 (COVID-19) in China: a report of 1014 cases. 200642 15. Hall LO, Paul R, Goldgof DB, Goldgof GM (2020) Finding covid-19 from chest x-rays using deep learning on a small dataset. arXiv preprint arXiv:2004.02060 16. Jin Y-H A rapid advice guideline for the diagnosis and treatment of 2019 novel coronavirus (2019-nCoV) infected pneumonia (standard version) Mil Med Res 2020 7 1 4 32029004 17. Hussain E Hasan M Rahman MA Lee I Tamanna T Parvez MZ CoroDet: a deep learning based classification for COVID-19 detection using chest X-ray images Chaos Solitons Fractals 2021 142 110495 10.1016/j.chaos.2020.110495 33250589 18. Ibrahim AU, Ozsoz M, Serte S, Al-Turjman F, Yakoi PS (2021) Pneumonia classification using deep learning from chest X-ray images during COVID-19. Cogn Comput 1–13 19. Zu ZY, et al (2020) Coronavirus disease 2019 (COVID-19): a perspective from China. 200490 20. Wang L Lin ZQ Wong A Covid-net: a tailored deep convolutional neural network design for detection of covid-19 cases from chest x-ray images Sci Rep 2020 10 1 1 12 31913322 21. Ahammed K, Satu MS, Abedin MZ, Rahaman MA, Islam SMS (2020) Early detection of coronavirus cases using chest x-ray images employing machine learning and deep learning approaches. medRxiv 22. Kibriya H, Rafique R, Ahmad W, Adnan S (2021) Tomato leaf disease detection using convolution neural network. In: 2021 international bhurban conference on applied sciences and technologies (IBCAST). IEEE, pp 346–351 23. Xu X A deep learning system to screen novel coronavirus disease 2019 pneumonia Engineering 2020 6 10 1122 1129 10.1016/j.eng.2020.04.010 32837749 24. Hemdan EE-D, Shouman MA, Karar ME (2020) Covidx-net: a framework of deep learning classifiers to diagnose covid-19 in x-ray images. arXiv preprint arXiv:2003.11055 25. Singh D Kumar V Kaur M Classification of COVID-19 patients from chest CT images using multi-objective differential evolution–based convolutional neural networks Eur J Clin Microbiol Infect Dis 2020 39 7 1379 1389 10.1007/s10096-020-03901-z 32337662 26. Loey M Smarandache F Khalifa NEM Within the lack of COVID-19 benchmark dataset: a novel gan with deep transfer learning for corona-virus detection in chest x-ray images Symmetry 2020 12 4 1 19 10.3390/sym12040651 27. Hernandez D, Pereira R, Georgevia P (2020) COVID-19 detection through x-ray chest images. In: 2020 international conference automatics and informatics (ICAI). IEEE, pp 1–5 28. Ismael AM Şengür A Deep learning approaches for COVID-19 detection based on chest x-ray images Expert Syst Appl 2021 164 114054 10.1016/j.eswa.2020.114054 33013005 29. Narin A, Kaya C, Pamuk ZJAPA (2020) Automatic detection of coronavirus disease (covid-19) using x-ray images and deep convolutional neural networks. 30. Wang S, et al. (2021) A deep learning algorithm using CT images to screen for Corona Virus Disease (COVID-19). Eur Radiol 1–9 31. Pathak Y, Shukla PK, Tiwari A, Stalin S, Singh S (2020) Deep transfer learning based classification model for COVID-19 disease. Irbm 32. Zhao J, Zhang Y, He X, Xie P (2020) Covid-ct-dataset: a ct scan dataset about covid-19. arXiv preprint arXiv:2003.13865 33. Bai X Predicting COVID-19 malignant progression with AI techniques SSRN J 2020 10.2139/ssrn.3557984 34. Bora DJ (2017) Importance of image enhancement techniques in color image segmentation: a comprehensive and comparative study. arXiv preprint arXiv:1708.05081 35. Hubel DH Wiesel TN Receptive fields, binocular interaction and functional architecture in the cat's visual cortex J Physiol 1962 160 1 106 154 10.1113/jphysiol.1962.sp006837 14449617 36. Marr D (1977) Representing visual information 37. Brownlee J (2019) A gentle introduction to the rectified linear unit (ReLU). Available: https://machinelearningmastery.com/rectified-linear-activation-function-for-deep-learning-neural-networks/ 38. Brownlee J (2019) A gentle introduction to pooling layers for convolutional neural networks. Available: https://machinelearningmastery.com/pooling-layers-for-convolutional-neural-networks/ 39. Hassanien AE Gaber T Handbook of research on machine learning innovations and trends 2017 Pennsylvania IGI global 40. Ouchicha C Ammor O Meknassi M CVDNet: a novel deep learning architecture for detection of coronavirus (Covid-19) from chest x-ray images Chaos Solitons Fractals 2020 140 110245 10.1016/j.chaos.2020.110245 32921934 41. (2020) Introduction to residual networks. Available: https://www.geeksforgeeks.org/introduction-to-residual-networks/ 42. He K, Zhang X, Ren S, Sun J (2016) Deep residual learning for image recognition. In: Proceedings of the IEEE conference on computer vision and pattern recognition. pp 770–778 43. Dietz M (2017) Understand deep residual networks—a simple, modular learning framework that has redefined state-of-the-art. Available: https://medium.com/@waya.ai/deep-residual-learning-9610bb62c355 44. Ramzan F A deep learning approach for automated diagnosis and multi-class classification of Alzheimer’s disease stages using resting-state fMRI and residual neural networks J Med Syst 2020 44 2 1 16 10.1007/s10916-019-1475-2 45. Abd ElGhany S Ibraheem MR Alruwaili M Elmogy M Diagnosis of various skin cancer lesions based on fine-tuned ResNet50 deep network CMC-Comput Mater Contin 2021 68 1 117 135 10.32604/cmc.2021.016102 46. Wu H Xin M Fang W Hu H-M Hu Z Multi-level feature network with multi-loss for person re-identification IEEE Access 2019 7 91052 91062 10.1109/ACCESS.2019.2927052 47. Suthaharan S Support vector machine Machine learning models and algorithms for big data classification 2016 New York Springer 207 235 48. Xue H Yang Q Chen S SVM: support vector machines The top ten algorithms in data mining 2009 Boca Raton Chapman and Hall/CRC 51 74 49. Kaggle (ed) COVID-19 radiography database 50. Kar P, Li S, Narasimhan H, Chawla S, Sebastiani F (2016) Online optimization methods for the quantification problem. In: Proceedings of the 22nd ACM SIGKDD international conference on knowledge discovery and data mining, pp 1625–1634. 51. Patel P, Kaggle (eds) (2021) Chest X-ray (Covid-19 & Pneumonia)

0

PMC9754310

NO-CC CODE

2022-12-16 23:26:35

no

Radiother Oncol. 2021 May 28; 158:S220-S221

latin-1

Radiother Oncol

10.1016/S0167-8140(21)06426-4

oa_other

==== Front Clean Technol Environ Policy Clean Technol Environ Policy Clean Technologies and Environmental Policy 1618-954X 1618-9558 Springer Berlin Heidelberg Berlin/Heidelberg 2456 10.1007/s10098-022-02456-1 Original Paper Exploration of CO2 emission reduction pathways: identification of influencing factors of CO2 emission and CO2 emission reduction potential of power industry Wang Weijun wwjhd@ncepu.edu.cn 1 http://orcid.org/0000-0002-9953-2247 Tang Qing tangqing@ncepu.edu.cn 1 Gao Bing 3012873@qq.com 2 1 grid.261049.8 0000 0004 0645 4572 Department of Economics and Management, North China Electric Power University, Baoding, 071003 Hebei China 2 Hengshui Power Supply Branch of State Grid Hebei Electric Power Co., Ltd, Hengshui, 053000 Hebei China 15 12 2022 115 30 3 2022 8 12 2022 © The Author(s), under exclusive licence to Springer-Verlag GmbH Germany, part of Springer Nature 2022, Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. Low-carbon development of China's power sector is the key to achieving carbon peaking and carbon neutrality goals. Based on the logarithmic mean divisor index (LMDI) model, considering the carbon transfer caused by inter-provincial electricity trading, this paper analyzes the influencing factors of CO2 emissions in the provincial power sector and uses K-means clustering method to divide 30 provinces into four categories to analyze the differences in regional carbon emission characteristics. In addition, by establishing different development scenarios, the carbon emission trends and emission reduction potentials of each cluster under different emission reduction measures from 2020 to 2040 are studied, in order to explore the differentiated emission reduction paths of each cluster. The results show that the contribution of influencing factors shows great differences in different provinces. Trends in CO2 emissions vary widely across scenarios. In the reference scenario, the CO2 emissions of each cluster will continue to increase; in the existing policy scenario, the total power industry will peak at 6.1Gt in 2030; in the advance peak scenario that puts more emphasis on the development of advanced technologies and renewable energy under the clean development model, the carbon emission peak will be brought forward to 2025, and the peak will be reduced to 5.2Gt. Finally, differentiated emission reduction paths and measures are proposed for the future low-carbon development of different cluster power industries, providing theoretical reference for the deployment of provincial-level emission reduction work, which is of great significance to the global green and low-carbon transformation. Graphical Abstract Supplementary Information The online version contains supplementary material available at 10.1007/s10098-022-02456-1. Keywords CO2 emissions Power industry Cluster analysis Emissions reduction potential ==== Body pmcIntroduction Global warming is a major problem faced by all mankind, and low-carbon emission reduction has become an important challenge for sustainable development in the future. CO2 is the largest supplier of the greenhouse effect and has become a key target of governance. Taking 2019 as an example, the world emits 36.4 billion tons of CO2, and China’s CO2 emissions are 10.2 billion tons, accounting for 28% of global emissions. As the world's largest emitter of greenhouse gases, China plays a vital role in mitigating global climate change. The combustion of fossil energy, especially coal, is the main source of CO2 emissions. In order to reduce carbon emissions, many countries restrict the use of coal and explore the supply of energy from alternatives to coal, such as wood (Jandačka et al. 2017), methane (Mardoyan and Braun 2014) and biofuels (Maroušek 2014). According to the World Energy Statistics Yearbook 2020, China's coal, natural gas and oil consumption accounted for 52, 8 and 15% of the world's total consumption, respectively, so it is clear that China's first priority to control CO2 emissions is to control coal consumption. About half of the coal in China is used to generate electricity, and 43% of China's total CO2 emissions in 2018 came from thermal power (about 90% of which is coal power), making it the largest source of CO2 emissions. However, China's resource endowment of “rich in coal, poor in oil and gas” makes it difficult for the power industry to leave coal in the short term. In the 14th Five-Year Plan Outline for ecological civilization construction (2021–2025), China proposed to resolutely implement the requirements of carbon peaking and carbon neutralization. Therefore, the power industry is facing the dual challenge of increasing supply and reducing emissions. Carbon peaking in the power industry is an essential prerequisite for China to achieve the carbon emission peaking target, and decarbonization of the power industry is the key to achieving a low-carbon society (Zhang et al. 2021). It should be pointed out that carbon emission characteristics of the power sector vary significantly between provinces (Wen and Li 2020). Therefore, it is critical to explore the carbon emission trajectory and peak time of the provincial power industry in order to formulate scientific and reasonable carbon emission reduction policies. The analysis of influencing factors has been used to study the changes of carbon emissions in various industry. The main methods to study influencing factors are SDA method (Luo et al. 2020), GDIM method (Yan et al. 2019a, b), IDA method (De Oliveira-De 2019) and Kaya method (Yang et al. 2020). In view of the significant regional and provincial differences, some literatures further discussed regional and provincial influencing factors. Specifically, Wang et al. (2018) analyzed the various influencing factors on the carbon emissions per unit of electricity from 1995 to 2014 in combination with the geographical differences of different regions. Based on the LMDI decomposition results, Liao et al. (2019) divided 30 provinces into five categories, and put forward emission reduction suggestions according to the characteristics of each category. Tian et al. (2021) used LMDI decomposition to analyze the differences in carbon intensity in different sectors and regions in China, and the results showed that energy and economic structure were the main factors for the differences in carbon intensity. Lin et al. (2019) quantified the impact of socioeconomic factors and population density on carbon emissions from the transportation industry based on the LMDI model, but ignored the transboundary issue of carbon dioxide generated during transportation. Based on the carbon emission data of the power sector from 2005 to 2016, He et al. (2020) constructed a spatial correlation network of carbon emissions in various provinces, and analyzed the spatial correlation characteristics and influencing factors. Many scholars have also studied the influencing factors in a single province or city, such as Shanghai (Wei et al. 2020), Baoding (Zhang et al. 2019), Beijing (Zhang et al. 2020), Shandong (Wang et al. 2017), Guangdong (Xu et al. 2021) and so on. Research on CO2 emission reduction potential and peak path in power sector has also achieved corresponding results. In the past 5 years, there have been many studies on the carbon emission reduction potential generated by the independent effects of factors such as the level of technology (He et al. 2021), carbon emission power structure (Cui et al. 2021) or carbon reduction policies (Chen and Chen 2019). Yang et al. (2021) set 3 different technology levels, and used the STRIPAT model to analyze the impact of technical factors on CO2 reduction in 6 sectors in China. Others referred to national and industry environmental protection policies, planning reports and technological development requirements to construct different scenarios for the possible economic structure, technical level, and energy consumption level in the future, and explored the potential for CO2 emission reduction under different scenarios (Yu et al. 2020). Yang et al. (2020) combined the LMDI method to decompose the influencing factors of CO2 emissions from 1996 to 2016 and discussed the influence degree of each factor. Considering the carbon sink technology, Demetriou and Hadjistassou (2021) developed four different energy structure scenarios and analyzed the reduction potential with a top-down approach. Carbon dioxide storage mostly refers to the high-pressure sealing of carbon dioxide into deep waste mines. The current cost of high-pressure storage in China is estimated to be 500 yuan per ton of CO2. However, the capture cost of biochar is relatively low, which is produced from natural biomass such as wood (Marousek and Gavurova 2022) or agricultural waste such as biogas (Marousek and Trakal 2022). It is currently an economically viable way to sequester carbon dioxide and can contribute to carbon emission reduction. The methods for forecasting peak carbon emissions mainly include the environmental Kuznets model (Jiang et al. 2019), scenario analysis (Wang et al. 2021) and IPAT model (Wen and Li 2019; Yin et al. 2022). Scholars used the above different methods to study the peak and peak time of carbon emissions in China, and drew different conclusions. Based on carbon emissions data of 8 sectors in China from 1995 to 2017, Fang et al. (2022) investigated the environmental Kuznets curve hypothesis of eight sectors, and predicted the peak time of each sector. Meng et al. (2017) used log-linear equations in a mixed model to predict the value of variables and set up five scenarios, and their results suggest that China's power sector will not reach the peak emission by 2030. Hernández and Fajardo (2021) set up three scenarios to estimate carbon emissions and carbon intensity in 2050 using the LEAP model. Based on the regional perspective, Tang et al. (2018) divided six regions based on geographical location, assessed the impact of technical factors and energy consumption on CO2 emissions from the regional power sector and studied the time of carbon peak. Chang et al. (2022) used scenario analysis to evaluate three carbon emission reduction scenarios from the perspectives of social equity, emission reduction efficiency and forest carbon sink. The results show that under the 2030 carbon emission target, the marginal CO2 emission reduction cost is 2315–5387 yuan. In summary, a lot of research has been carried out on the influencing factors, emission reduction potential and peak size of carbon emissions in China, but there are still the following shortcomings in the current studies: (1) There are many studies that analyze the influencing factors of carbon emissions in the power industry from the national level or a single region, province and city, while there are fewer studies that explore regional differences in each region. (2) The existing literature analyzes the regional CO2 emissions mainly according to the geographical location or the surface characteristics, ignoring the potential characteristics of carbon emissions. (3) There are more analyses of the peak time and peak size of the entire power industry, while only a few scholars have studied the carbon peak paths of different clusters based on the carbon emission characteristics of the power industry. Under the background of climate change, how to put forward targeted measures to promote the peak value according to the characteristics of regional carbon emissions and avoid the "flooding" of energy-saving and emission-reduction policies is an urgent problem to be solved. In response to previous research deficiencies, this paper adopts LMDI method to quantitatively analyze the influence degree of main factors of China's power industry from 2000 to 2019. Combined with the decomposed influencing factors, it is hypothesized that there are regional differences in the influencing factors of carbon emissions, and cluster analysis is carried out on 30 provinces according to the decomposition results from 2015 to 2019. According to China's medium and long-term economic development and energy demand, the change of carbon emissions from power production in 2020–2040 is divided into 3 scenarios: reference scenario, existing policy scenario and advance peak scenario. The carbon emission trends of the whole power industry and different clusters under different scenarios are predicted, and the emission reduction potential is analyzed. Based on the scenario analysis results, the contribution of each factor to future CO2 emission reduction is analyzed. Finally, targeted measures are made to promote peak attainment in the power sector. It can provide a reference for policy makers to formulation of provincial carbon emission reduction policies, which is of great significance to global climate change. Methods and dataset Estimation of CO2 emissions from power industry In the power sector, the consumption of fossil energy is a major contributor to carbon emissions. Accordingly, this paper selects a total of 22 fossil energy sources, including coal, petroleum and natural gas. The carbon emissions of the power industry are measured through the measurement model proposed by the United Nations Intergovernmental Panel on Climate Change in 2006 (Yan et al. 2019a, b). The fuel parameters and emission coefficients involved are shown in Table 1, and the calculation formula is as follows:1 C=∑i=13Ci=∑i=13Ei×EFi=∑i=13Ei×NCVi×CEFi×COFi×44/12 Table 1 Fuel parameters and emission factors Energy name Average low calorific valuea (kJ/kg or kJ/m3) Conversion coefficient of standard coala (kg/kg or kg/m3) Carbon content per unit calorific valueb (tC/TJ) Carbon oxidation rateb (%) Raw coal 20,934 0.7143 26.37 0.94 Washed coal 26,377 0.9000 25.41 0.93 Other coal washing 8374 0.4286 25.41 0.93 Coal products 20,908 0.6000 33.60 0.90 Gangue 8374 0.2857 25.80 0.93 Coke 28,470 0.9714 29.50 0.93 Coke oven gas 18,003 0.6143 13.58 0.99 Blast furnace gas 3768 0.1286 70.80 0.99 Converter gas 7945 0.2714 49.60 0.99 Other gas 5227 0.3571 12.20 0.99 Other coking products 28,435 1.3000 29.50 0.93 Crude oil 41,868 1.4286 20.10 0.98 Gasoline 43,124 1.4714 18.90 0.98 Kerosene 43,124 1.4714 19.60 0.98 Diesel oil 42,705 1.4571 20.20 0.98 Fuel oil 41,868 1.4286 21.10 0.98 Petroleum coke 31,947 1.0918 27.50 0.98 Liquefied petroleum gas 50,242 1.7143 17.20 0.98 Refinery Gas 46,055 1.5714 18.20 0.98 Other petroleum products 41,031 1.4000 20.00 0.98 Natural gas 38,979 1.3300 15.30 0.99 Liquified natural gas 51,498 1.7572 17.20 0.98 aData are collected from China Energy Statistical Yearbook in 2019 and General Principles for Calculation of Comprehensive Energy Consumption GB/T 2589–2020 bData are collected from Guidelines for Compilation of Provincial Greenhouse Gas Inventory where C is CO2 emission, i is the energy type, Ei is the consumption of energy i (physical quantity), EFi is the carbon emission factor of energy i, NCVi is the average low heating value of energy i, CEFi denotes the carbon content per unit heat generated of energy i, COFi denotes the carbon oxidation rate of energy i, and 44/12 is the molecular weight ratio of CO2. LMDI decomposition method LMDI decomposition method is widely used in the analysis of influencing factors, which has the characteristics of residual-free and wide applicability (Jiang et al. 2020). This paper establishes a decomposition model based on LMDI method to analyze the influencing factors of CO2 emission change in the power sector. The change of CO2 emission generated by power generation can be decomposed into:2 C=∑i=13Ci=∑i=13CiE×ETP×TPEG×EGPD×PD=∑i=13FSi×EI×PS×ET×PD where Ci is the CO2 emission of the energy type i, E is the fossil energy consumption, TP is the thermal power generation, EG is the total power generation, and PD is the total power demand. FSi, EI, PS and ET are represent the fuel structure, energy efficiency, power generation structure, and electricity trade, respectively. The decomposition formula for the change in CO2 emissions can be described as follows:3 ΔC=CiT-Cit=ΔCFS+ΔCEI+ΔCPS+ΔCET+ΔCPD where ΔCFS, ΔCEI, ΔCPS, ΔCET and ΔCPD are the changes of FS, EI, PS, ET, and PD that affect the CO2 emission change from year t to year T, respectively, and CiT and Cit are the CO2 emission of energy type i in year T and t, respectively. ΔCFS, ΔCEI, ΔCPS, ΔCET and ΔCPD are calculated as follows, where w is the log mean weight:4 ΔCFS=∑i=13w×lnFSiTFSit 5 ΔCEI=∑i=13w×lnEITEIt 6 ΔCPS=∑i=13w×lnPSTPSt 7 ΔCET=∑i=13w×lnETTETt 8 ΔCPD=∑i=13w×lnPDTPDt 9 w=CiT-CitlnCiT/Cit K-means clustering analysis method The K-means clustering algorithm is based on Euclidean distance and is one of the most common statistical clustering methods. It can be clustered according to the spatial position of each object to be clustered, and the individuals with similar distances, that is, with similar characteristics, can be clustered into one class. First define the number of clusters k, and then assign the dataset into k clusters. Calculate the distance of the data point to the cluster center and redistribute the dataset into the closest cluster. And repeat the above steps through multiple iterations, and finally get the result of clustering. Scenario analysis Scenario analysis is a commonly used multivariate forecasting method to study the possible outcomes of a combination of factors. Scenario analysis has been widely used in carbon emission projections in recent years, which can predict the changing trend of carbon emission in the future. Through the multi-scenario setting, that is, simulating different paths of future development, the corresponding variable change speed parameters can be formulated, and the carbon emission trend under different scenarios can be predicted, and the CO2 emission reduction potential can be obtained through comparative analysis. Data sources In this paper, the physical quantities of all kinds of fossil fuels in 30 provinces from 2000 to 2019 are obtained from the China Energy Statistical Yearbook. In addition, the total power generation, thermal power generation and power consumption of each province from 2000 to 2019 are from the China Statistical Yearbook. The average low calorific value and reduced standard coal coefficient of fossil fuels come from China Energy Statistical Yearbook in 2019 and General Principles for Calculation of Comprehensive Energy Consumption GB/T 2589–2020, and the carbon content and carbon oxidation rate of unit calorific value come from Guidelines for Compilation of Provincial Greenhouse Gas Inventory. It is assumed that the fuel parameters and emission coefficient are constant in the time span analyzed in this paper. Results and discussion Historical carbon emission calculation results Figure 1 shows the change in CO2 emissions from the power sector from 2000 to 2019, showing an overall upward trend, from 1.0 Gt in 2000 to 4.8Gt in 2019.Fig. 1 CO2 emissions in China's power industry from 2000 to 2019 From 2000 to 2010, total CO2 emissions increased from 1.0 Gt to 3.2 Gt, and most of the growth rates were above 10% and fluctuating. The reason behind this is that in the initial stage of the electricity market reform, the cost advantage of coal power was obvious, far lower than the cost of clean energy power generation, which promoted the growth of CO2 emissions. From 2011 to 2019, under the background of economic development, the power industry developed rapidly and production capacity expanded rapidly. However, driven by measures such as energy emission control, the growth rate of total emissions slowed down and was basically controlled at around 5%, from 3.7 to 4.5 Gt. In general, since the fuel structure was dominated by coal, the CO2 emissions from power industry mainly came from coal-fired power generation, accounting for 98.4% of the total. The proportion of CO2 emissions from oil-fired power generation showed an overall downward trend, falling to 0.20% of the total by 2019. The proportion of CO2 emissions from gas-fired generation has been increasing, reaching 2.01% of the total by 2019 and surpassing oil-fired generation. Analysis of decomposition results Based on the LMDI method, the CO2 emission factors of the power industry from 2000 to 2019 are decomposed. Figure 2 shows the contribution of each influencing factor to carbon emissions. The period from 2000 to 2019 is subdivided into four time periods, and the decomposition results of these four periods are analyzed in depth in this paper.Fig. 2 LMDI decomposition results of influencing factors from 2000 to 2019 Between 2000 and 2005, the industry's CO2 emissions increased by 1.0Gt, with a growth rate of 98.10%. Energy efficiency effect, electricity trade effect and power demand scale effect contributed 98.58% increase in total CO2 emissions, which was slightly offset by changes in fuel structure (− 0.43%) and power generation structure (− 0.05%). Between 2005 and 2010, emissions increased by 1.1Gt, with a growth rate of 55.82%. The power demand had a reduced impact on CO2 emissions (63.63%). During this period, the energy efficiency and power generation structure were optimized, and the impact on carbon emission reduction increased by − 9.77 and 4.21%, respectively. From 2010 to 2015, CO2 emissions of the industry increased by 0.5Gt, a growth rate of 16.36%. The impact of the electricity demand was further reduced (32.57%), and coupled with the fuel structure effect and the electricity trade effect, the positive influencing factors led to a 37.77% increase in total. The impact of energy efficiency and power generation structure effect on carbon emission reduction continued to increase, which together offset the impact of 21.42%. Between 2015 and 2019, CO2 emissions increased by 0.8Gt, with a growth rate of 21.71%. FSi, EI, PS, ET and PD contributed 3.20, − 3.31, − 6.30, 1.79 and 26.34%, respectively. The results show that the electricity demand factor has dominated CO2 emissions over the past 20 years, however the contribution is gradually decreasing as more of the new electricity demand is met by clean energy. The most critical factor in reducing CO2 emissions was the improved of energy efficiency, followed by the optimization of power generation structure. Since the coal-dominated fuel mix has not been improved, the fuel mix effect has a very limited impact on carbon emissions. Changes in fuel structure and electricity trade also led to a slight increase. Figure 3 shows the contribution of each province's influencing factors to carbon emissions. It can be seen that the contribution of each province's carbon emission influencing factors to carbon emissions varies greatly and changes dynamically over time, which verifies the research hypothesis of this paper. Except for the power demand scale effect that promoted carbon emissions in all provinces, other influencing factors had positive or negative effects on CO2 emissions in all provinces. Between 2000 and 2005, the main driver of provincial CO2 emissions was power demand scale effects. Electricity trade caused inter-provincial carbon transfer, in some provinces affected by “north–south power supply” and “west–east power transmission”, the power trade effect played a key role. For example, Guangdong, Shandong, Shanghai, Hebei and other power outsourcing provinces have thus reduced part of CO2 emissions, while Guizhou, Yunnan, Shanxi, Shaanxi and other power outsourcing provinces have increased CO2 emissions. The power generation structure effect reduced CO2 emissions in Sichuan, Hubei, Zhejiang and other provinces with high level of clean energy utilization. Energy efficiency significantly contributed to Shandong's CO2 emissions (6.30%). The impact of fuel structure on CO2 emissions in each province, whether positive or negative, was quite limited.Fig. 3 Relative effects of the influencing factors by province from 2000 to 2019 During the period from 2005 to 2010, the power demand effect remained the major factor driving CO2 emissions in various provinces. Jiangsu's power demand effect was the most significant, resulting in the national emission growth of 11.00%. After the implementation of the policy of eliminating inefficient generating units (Qin et al. 2020), the energy efficiency effect and power generation structure effect reduced part of CO2 emissions, contributing − 18.45 and − 5.80% to national emissions, respectively. The influence of power trade continued to increase in the provinces with “north–south power supply” and “west–east power transmission”. Due to the continuous increase in power demand, a large amount of coal-fired units had been invested, which increased the impact of fuel structure effects on CO2 emissions in various provinces, with a contribution rate of 8.77%. Between 2010 and 2015, the contribution of the power demand effect to provincial CO2 emissions further increased. Shandong's power demand effect was considered the most significant driver, contributing 31.70% of the national emissions growth. The energy efficiency effect made a significant contribution to the emission reduction, with Shandong and Hebei offsetting 10.19 and 8.79% of the national CO2 emissions, respectively. Power generation structure effect (Sichuan, etc.), fuel structure effect (Guizhou, etc.) and electricity trade effect (Henan, etc.) were the key factors of CO2 emission reduction in some provinces. From 2015 to 2019, among provinces, the power demand effect had the largest pulling effect on CO2 emissions. For example, the contribution rate of power demand effect in Inner Mongolia and Shandong Province to the national CO2 emissions is 16.36 and 11.47%, respectively. The power demand effect was offset by the energy efficiency effect in most provinces (Jiangsu, etc.) and the power generation structure effect in most provinces (Shandong, etc.), reducing the national CO2 emissions by −21.73 and − 27.90%, respectively. The carbon emission reduction of Zhejiang (− 2.45%), Shaanxi (–2.06%) and Jiangsu (− 1.43%) were significantly affected by electricity trade effect. Fuel structure promoted carbon emissions in Jiangsu, Hebei, Beijing and Guangdong. In general, the influencing forces varied by province, showing different effects across the four phased studies. Therefore, carbon reduction strategies should vary according to the specific conditions of each region. Analysis of inter-provincial clustering and regression results Cluster analysis Faced with severe air pollution, the Chinese government has put forward a series of emission reduction measures, but there is still a big gap from the goals of carbon peak and carbon neutralization. Compared with the carbon emission reduction plans of other countries, such as the implementation of carbon sink projects in British agriculture and animal husbandry, the implementation of green building concept in Germany, and the development of a complete carbon tax system in Finland, the low-carbon transformation of China's power industry needs more targeted emission reduction plans. Based on the K-means clustering method, taking 5 influencing factors and CO2 emissions of provinces from 2015 to 2019 as clustering variables, 30 provinces in China are divided into four clusters, in order to obtain more targeted emission reduction strategies. Figure 4 shows the clustering results of provinces, and Fig. 5 shows the carbon emission characteristics of different clusters.Fig. 4 China's provincial classification and influencing factors effect from 2015 to 2019 Fig. 5 Influencing factors effect of each cluster from 2015 to 2019 Cluster 1 contributed 27.17% to the growth of carbon emissions, so the low-carbon transition of cluster 1 is urgent. Different from other clusters, all influencing factors of cluster 1 were positively pulling CO2 emissions, in which the expansion of power demand scale played a leading role, and power generation structure and electricity trade also promoted CO2 emissions. It is suggested to eliminate inefficient coal-fired power plants or introduce advanced technology to gradually improve fuel efficiency and optimize power structure. Among the four clusters, cluster 2 caused the least increase in carbon emissions, accounting for only 4.30% of the national emission increment from 2015 to 2019. Growth in electricity demand contributed most to the cluster's CO2 emissions (13.14%), but was largely offset by energy efficiency effects (− 7.81%) and generation mix effects (− 8.10%). Cluster 3 covers four provinces, which contributed to a smaller increase in national carbon emissions, accounting for 9.35%. Different from other clusters, the effect of electricity trade on provinces in this cluster played a promoting role in carbon emission reduction. Having to purchase electricity from outside because local power generation could not meet demand, which resulted in carbon leakage and thus reduced CO2 emissions in the provinces. Cluster 4 contributed 59.18% from 2015 to 2019, resulting in the largest increase, so cluster 4 is the key area for emission reduction. The expansion of power demand played a leading role (69.57%) in the CO2 emissions of provinces in the cluster, and the change of fuel structure also promoted CO2 emissions. Energy efficiency and power generation structure offset a total of 20.82% of the increase. In the cluster, Jiangsu, Shandong and Hebei are major industrial provinces, and the high proportion of heavy industries and the high coal consumption are the fundamental reasons for the large contribution of these provinces to emissions. Therefore, provinces in cluster 4 need to introduce more advanced production technologies, improve industrial efficiency and encourage the development of new technology industries to achieve a low-carbon transition. In terms of emission reduction measures, the economic feasibility of implementing different clusters varies. From the perspective of economic feasibility, compared with cluster 3 and cluster 4, most of the provinces in cluster 1 and cluster 2 are economically underdeveloped areas, which means that the implementation of carbon emission reduction policies in the power industry in these two clusters is more challenging and may face economic pressure. Therefore, the region should provide appropriate financial incentives, such as subsidies for the introduction of new technologies, which can provide targeted financial assistance for the promotion of carbon emission reduction policies in the power industry. Fitting analysis of cluster carbon emissions China's unique resource endowment has prompted China's energy consumption to be heavily biased toward coal. The proportion of coal consumption represents China's energy structure to a certain extent, and the proportion of coal input in the power generation process greatly affects the cleanliness of production. Therefore, in this paper, the proportion of coal consumption is used to characterize the structure the fuel structure in the actual research. The formula is obtained:10 C=f(PC,EI,PS,ET,PD) The logarithm of the formula can be obtained:11 lnC=α1lnPC+α2lnEI+α3lnPS+α4lnET+α5lnPD+α0 where, PC represents the proportion of coal and its products in fossil energy consumption; EI represents energy efficiency; PS represents power generation structure; ET and PD represent electricity trade and power demand, respectively. Firstly, time series data of cluster 1 are tested by unit root test, and as shown in Table 2 *** and ** denote the original assumed significance level of the existence of the unit root at a point is 1 and 5%, respectively. As can be seen from the test results, not all variables are stable. Each variable is integrated of order 2 and can be cointegration test. In this paper, the Engle-Granger two-step method is used to conduct cointegration test. Firstly, regression is performed for variables, and then unit root test is performed for residual items. If it is stable, it indicates that there is a cointegration relationship between variables. The stability test results of the residual term are shown in Table 3. It can be seen from the results that the residual term is stable and there is a co-integration relationship between variables, thus obtaining the co-integration equation:12 lnC1=-0.387lnPC+0.961lnEI+0.943lnPS+0.698lnET+1.072lnPD-1.952 Table 2 Results of unit root test Variables ADF test P–P test Variables ADF test P–P test Variables ADF test P–P test lnC −2.029 −2.423 d(lnC) −3.34** −3.341** d(lnC,2) −5.70*** −12.1*** lnPC −3.944** −3.943** d(lnPC) −7.26*** −16.3*** d(lnPC,2) −9.07*** −23.4*** lnEC −0.925 −0.803 d(lnEC) −4.97*** −5.18*** d(lnEC,2) −7.17*** −18.4*** lnPS −3.217** −3.217** d(lnPS) −5.93*** −5.93*** d(lnPS,2) −9.58*** −23.0*** lnET −2.114 −2.115 d(lnET) −4.32** −4.32** d(lnET,2) −7.90*** −10.9*** lnPD −3.60** −3.25** d(lnPD) −2.049 −1.979 d(lnPD,2) −4.84*** −8.39*** Table 3 Stationary test of residual term Residual term ADF test P–P test Resid01  − 5.356***  − 5.919*** Similarly, the cointegration equations of clusters 2, 3 and 4 can be obtained:13 lnC2=-3.327lnPC+0.858lnEI+0.932lnPS+1.073lnET+1.007lnPD-1.522 14 lnC3=-0.243lnPC+0.869lnEI+0.645lnPS+0.786lnET+0.946lnPD-1.032 15 lnC4=-9.399lnPC+0.868lnEI+0.573lnPS+0.801lnET+1.054lnPD-1.965 The following conclusions can be drawn from the equations for each of the above clusters: All the coefficients in the cointegration equation of each cluster are different, which further verifies the research hypothesis that there are regional differences in the influencing factors of carbon emissions. The synergistic efficiencies of ln EC, ln PS, ln ET and ln PD of the four clusters are all positive, while the coefficient of lnPC is negative, which is consistent with the social and economic development. The coefficients of coal consumption ratio are − 0.387, − 3.327, − 0.243 and − 9.399 in the four clusters, respectively, indicating that the relationship with carbon emissions is negative. In addition, the coefficients of cluster 1 and cluster 3 are larger, which is due to the fact that the fuel structure of these two clusters has not been adjusted significantly in the past 20 years, resulting in that the carbon emission level is less affected by the proportion of coal. The proportion coefficient of coal in cluster 4 is the smallest, which indicates that the change of fuel structure in this cluster has the greatest impact on carbon emissions, and the optimization and adjustment of fuel structure will be conducive to the carbon emission reduction of this cluster. However, most of the provinces in this cluster are coal power provinces, which means that it is difficult to change the current situation that coal power occupies the dominant position in a short time. The coefficient of energy efficiency and power generation structure is positive, indicating that improving energy efficiency and optimizing power generation structure will significantly improve carbon emission reduction potential. Cluster 1 has the largest coefficient of these two factors, indicating that the cluster can effectively promote carbon emission reduction by shutting down inefficient units, introducing advanced technologies, and vigorously developing renewable energy power generation. The generation structure coefficient of cluster 4 is the smallest, so the power generation structure of cluster 4 needs to be vigorously adjusted to play a role in carbon emission reduction. The coefficient of electricity trade and power demand is positive, indicating that carbon leakage caused by inter-provincial and regional power exchange will increase the carbon emission of power output province, and the expansion of the scale of electricity demand will also promote the increase of carbon emissions. Cluster 2 has more power exporting provinces, and its electricity trade coefficient is the largest, indicating that inter-provincial and regional power exchange will lead to the increase of carbon emission of the provinces in the cluster. The newly increased power demand of cluster 3 is more satisfied by clean energy, resulting in that the impact of power demand on CO2 emissions of this cluster less than that of other clusters, and the power demand coefficient of cluster 3 is the minimum. Therefore, it is recommended to promote power substitution in this cluster. Analysis of CO2 emission reduction potential Scenario hypothesis In order to clarify the carbon emission trend of China's power industry before 2040, this paper sets up three different development scenarios, namely reference scenario, existing policy scenario and advance peak scenario. Each scenario represents a corresponding future development policy planning path of China's power industry. It should be noted that in 2020, due to the impact of the COVID-19, the expansion of power demand scale slows down, and the electricity demand growth rate for that year is set at 3.1% for each cluster in all 3 scenarios. The reference scenario assumes that China's power sector will not take any additional carbon reduction measures from 2019. In the reference scenario, the demand for electricity continues to grow with the development of the economy. From 2021 to 2040, the average annual growth rate of terminal power consumption of the four clusters will reach 6.85, 4.25, 6.93 and 6.61%, respectively. Power generation structure, the share of coal in fuels, energy efficiency and electricity trade will remain unchanged at 2019 levels. The existing policy scenario are set based on the future development policies and planning documents promulgated by China’s power industry, including the 14th Five-Year Plan for Energy Development, the 14th Five-Year Plan for Renewable Energy Development, and the 2035 Visionary Goals Outline. Considering the increasingly severe resource and environmental pressures, China should actively take measures such as demand-side management to control energy consumption during this period. And gradually increase the proportion of low-carbon energy power generation, constantly eliminate backward coal power units and adding high-efficiency coal power units. The advance peak scenario uses more aggressive policies than the existing policy scenario to accelerate the low-carbon transition in the power sector, and would set stricter targets for the control of total energy consumption, which is a relatively optimistic low carbon development scenario. The progress of industrial structure will continue to optimize the power demand structure, and the proportion of electricity used by the tertiary industry and residents will continue to rise. Compared with existing policy scenarios, the proportion of low-carbon energy generation in China will be further improved, and a part of small thermal power will guarantee power supply in the regional grid. To explore the carbon reduction potential of power sector, this paper first predicts the future trend of the variables in Eqs. (12)-(15). This paper sets the parameters of coal proportion, energy efficiency, power generation structure, electricity trade and power demand scale in China from 2020 to 2040 by referring to the optimal scenario design of power decoupling by Wang et al. (2021) and Dong et al. (2021), and the parameter setting results are shown in Table 4.Table 4 Scenario parameters of China's power industry by cluster from 2020 to 2040 Reference scenario Existing policy scenario Advance peak scenario 2025 2030 2035 2040 2025 2030 2035 2040 2025 2030 2035 2040 Cluster 1 PC(%) 0.994 0.994 0.994 0.994 0.994 0.973 0.949 0.903 0.964 0.927 0.904 0.881 EI(%) 0.293 0.293 0.293 0.293 0.291 0.285 0.283 0.277 0.288 0.281 0.279 0.272 PS(%) 0.608 0.608 0.608 0.608 0.608 0.578 0.522 0.472 0.572 0.491 0.401 0.327 ET(%) 1.129 1.129 1.129 1.129 1.135 1.146 0.151 1.158 1.079 1.134 1.135 1.135 PD(100 billion kWh) 1.817 2.531 3.524 4.908 1.557 1.859 1.953 2.053 1.666 1.839 2.030 2.242 Cluster 2 PC(%) 0.994 0.994 0.994 0.994 0.994 0.994 0.984 0.974 0.995 0.995 0.993 0.988 EI(%) 0.282 0.282 0.282 0.282 0.279 0.276 0.275 0.273 0.276 0.274 0.273 0.272 PS(%) 0.479 0.479 0.479 0.479 0.451 0.428 0.387 0.332 0.446 0.403 0.361 0.310 ET(%) 1.056 1.056 1.056 1.056 1.062 1.072 1.078 1.083 1.025 1.040 1.046 1.051 PD(100 billion kWh) 2.167 2.669 3.287 4.047 2.070 2.373 2.495 2.662 2.160 2.326 2.506 2.700 Cluster 3 PC(%) 0.994 0.994 0.994 0.994 0.982 0.972 0.953 0.934 0.976 0.952 0.928 0.905 EI(%) 0.261 0.261 0.261 0.261 0.259 0.258 0.256 0.253 0.259 0.257 0.254 0.252 PS(%) 0.721 0.721 0.721 0.721 0.639 0.578 0.496 0.426 0.601 0.490 0.399 0.309 ET(%) 0.814 0.814 0.814 0.814 0.839 0.851 0.864 0.877 0.790 0.794 0.798 0.802 PD(100 billion kWh) 2.276 3.181 4.446 6.215 1.944 2.319 2.438 2.562 2.059 2.273 2.510 2.638 Cluster 4 PC(%) 0.999 0.999 0.999 0.999 0.998 0.997 0.996 0.995 0.998 0.998 0.997 0.996 EI(%) 0.304 0.304 0.304 0.304 0.302 0.300 0.297 0.294 0.302 0.299 0.296 0.293 PS(%) 0.860 0.860 0.860 0.860 0.762 0.689 0.573 0.467 0.762 0.590 0.456 0.353 ET(%) 1.097 1.097 1.097 1.097 1.130 1.147 1.165 1.182 1.033 1.038 1.043 1.048 PD(100 billion kWh) 3.855 5.310 7.314 10.07 3.340 3.986 4.294 4.626 3.451 3.885 4.289 4.736 Analysis of CO2 emission reduction potential By substituting the parameters of each scenario into Eqs. (12)–(15), the CO2 emissions of each cluster and the whole country before 2040 under each scenario are obtained as shown in Fig. 6. In general, the CO2 emission trends vary widely among scenarios due to the different intensity of policy measures adopted.Fig. 6 Prediction and comparison of CO2 emissions from 2000 to 2040 under different scenarios of each cluster Under the reference scenario, CO2 emissions of each cluster will continue to increase. Due to the absence of active policies to combat climate change, the demand for electricity will rise significantly, but the fuel structure, energy efficiency and power generation structure will not be improved, and the technological level is stagnant, leading to the CO2 emissions of cluster 1–4 to 2040 reaching 3.2, 1.8, 2.6 and 9.3 Gt, respectively, and the total national emissions are 1.7 Gt. This indicates that carbon emissions from China's power sector will not peak before 2030 if measures are not taken. Under the existing policy scenario, according to the promulgated energy plan, the power industry will significantly reduce CO2 emissions, and the CO2 emission in 2040 will be reduced by 1.1 Gt compared with the reference scenario. The entire industry will peak at 6.1 Gt in 2030, and carbon emissions will gradually decrease from 2030. The four clusters will peak in 2030, 2029, 2030 and 2032, respectively, with the peak values of 1.1, 0.9, 0.9 and 3.2Gt. In the advance peak scenario, more stringent power demand side management measures are implemented, and the power supply structure is further improved. China's power sector will peak at 5.2Gt in 2025, 5 years earlier than the existing policy scenario, and with 0.9 Gt lower peak carbon emissions. CO2 emissions of each cluster will peak at 0.9, 0.8, 0.8 and 2.7 Gt in 2025, 2024, 2025 and 2027, respectively. Compared with the existing policy scenario, the total CO2 emissions in 2040 will be reduced by 20.52% under the advance scenario. Analysis on contribution of influencing factors of CO2 emission In order to understand the contribution of influencing factors to CO2 emission in each cluster under the four scenarios, the LMDI model constructed above is used to decompose carbon emissions in each scenario, and the emission reduction contribution of each influencing factor is shown in Table 5.Fuel structure (ΔFS) In the existing policy scenario, the total contribution of four clusters’ fuel structure optimization to CO2 emissions is 1.1Gt, among which cluster 4 contributes the largest amount of 0.8Gt; in the advance peak scenario, the total contribution of four clusters to CO2 emissions is 1.4Gt, among which cluster 4 contributes the largest amount of 1081.56Mt. In the future, with the improvement and promotion of carbon capture technology and the installation of CCUS equipment in coal-fired units, carbon emissions will be further reduced, which will play a more important role in CO2 emission reduction. Energy efficiency (ΔEI) Improvements in thermal power generation efficiency will continue to be one of the most important factors contributing to CO2 reductions. From 2019 to 2040, in the existing policy scenario and the earlier peak scenario, the improvement of thermal power generation energy efficiency in the four clusters contributes 0.2 and 0.2Gt to CO2 emission reduction, respectively. However, as the efficiency of power generation continues to improve in the future, the rate of technological progress will gradually decrease. Meanwhile, due to the existence of carbon lock-in effect of thermal power generation, the emission reduction contribution of energy efficiency may gradually decrease. Power generation structure (ΔPS) In the existing policy scenarios, the contribution of low-carbon energy generation to CO2 emissions in the four clusters reached − 0.2, − 0.3, − 0.4, and − 1.6Gt, respectively, contributing − 17.08, − 22.91, − 32.30, and − 130.05%, far exceeding 4.34, − 8.10, − 6.52, − 17.61% in 2000–2019; in the advance peak scenario, due to the more emphasis on low-carbon energy development, the impact of power generation structure to CO2 emissions reaches − 0.4, − 0.3, − 0.6 and − 2.1Gt, which is the most effective CO2 emission reduction factor. With the problems of traditional energy depletion and increased pressure on environmental protection, China will accelerate the development of renewable energy generation, and the emission reduction effect of low-carbon energy power generation will be further highlighted. Electricity trade (ΔET) In terms of relative quantity, the contribution of power trade effect to CO2 emission of power industry is small. In the existing policy scenario, from 2019 to 2040, the power trade effect of the four clusters contributed 1.70, 1.58, 4.59 and 15.97%, respectively. However, in terms of absolute amount, the power trade effect also makes a great contribution to emission reduction. The cumulative CO2 emissions from the electricity trading effect for the power sector in 2019–2040 amount to 0.3Gt and − 0.1Gt in the existing policy scenario and the advance peak scenario, respectively. In the future, with the completion of digital power grid and modern power grid, the power trade effect will make a continuous and stable contribution to CO2 emission reduction. Power demand (ΔPD) According to the decomposition results, from 2019 to 2040, the factor that contributes the most to CO2 emissions in each scenario is the terminal power demand. From 2019 to 2040, the power demand effect under the three scenarios will increase by 0.1, 0.3 and 2.4Gt of CO2, respectively. This means that the continuous guidance of electricity conservation through power demand-side management and active promotion of energy-saving products will become one of the most effective means of reducing CO2 emissions in the power industry in the future. Table 5 Analysis of contribution of influencing factors from 2019 to 2040 Scenario ΔPC(Gt) ΔEI(Gt) ΔPS(Gt) ΔET(Gt) ΔPD(Gt) ΔC(Gt) Reference scenario Cluster 1 0.000 0.000 0.000 0.000 2.4 2.4 Cluster 2 0.000 0.000 0.000 0.000 1.0 1.0 Cluster 3 0.000 0.000 0.000 0.000 1.9 1.9 Cluster 4 0.000 0.000 0.000 0.000 7.0 7.0 Existing policy scenario Cluster 1 0.08 −0.05 −0.2 0.2 0.4 0.2 Cluster 2 0.08 −0.2 −0.3 0.02 0.3 0.1 Cluster 3 0.1 −0.02 −0.4 0.05 0.4 0.1 Cluster 4 0.8 −0.08 −1.6 0.2 1.4 0.8 Advance peak scenario Cluster 1 0.08 −0.06 −0.4 0.01 0.4 0.01 Cluster 2 0.04 −0.03 −0.3 −0.004 0.3 0.03 Cluster 3 0.2 −0.02 −0.6 −0.01 0.4 −0.05 Cluster 4 1.1 −0.09 −2.1 −0.1 1.3 0.09 Drawing from the above, the decomposition results of the factors affecting carbon emissions in China's power industry are fuel structure, energy efficiency, power generation structure, power trade and power demand. Compared with the LMDI decomposition results with socioeconomic indicators as the main influencing factors, this paper pays more attention to the direct influencing factors of carbon emissions in the power industry, and considers the impact of carbon leakage caused by power trade in various regions, which is helpful to formulate targeted emission reduction policies. According to the analysis results of influencing factors, power generation structure and energy efficiency are the two most important factors for the low-carbon transformation of the power industry. In the scenario analysis, the emission reduction measures under different scenarios have a significant impact on CO2 emissions. Due to the active policies adopted in fuel structure, energy efficiency, power generation structure, power trade and other aspects, compared with the reference scenario, the entire power industry will peak in 2030 and 2025 under the existing policy scenario and the advance peak scenario, respectively, and the peak value under the advance scenario is lower. The results show that exploring differentiated emission reduction paths in different clusters will help achieve the goal of peaking the power industry by 2030. This study has some limitations that are worth noting. This paper focuses on the carbon emission analysis of China's power production, and does not consider the carbon emissions of power transmission and use. If there are reliable data, this point deserves attention in the future. In future research, it is necessary to analyze the carbon emissions in the whole process of power industry from production, transportation to use. In addition, the economic cost analysis of policy implementation is also the direction that needs to be worked on in the future. Researchers can also use other methods and samples (data of power industry in other countries, even data of other industries) to obtain new insights. Conclusions and recommendations This study decomposes the influencing factors of carbon emissions in China's power industry, verifies the research hypothesis that there are regional differences in the influencing factors of carbon emissions through cluster analysis and carbon emission regression model fitting analysis, and finally discusses the carbon emission trend and emission reduction potential of the power industry under different development modes through scenario setting, revealing the following results. First, different cluster carbon emission models (Eqs. 12–15) confirm the long-term relationship between various influencing factors and CO2 emissions, which indicates that carbon emission reduction policies should be differentiated based on regional characteristics. Secondly, the active countermeasures in the scenario analysis are effective in advancing the peak time of carbon emissions and reducing the peak of carbon emissions. Compared with the reference scenario, the current policy scenario and the advance peak scenario for the entire power industry will reach the peak value in 2030 (6.1Gt) and 2025 (5.2Gt), respectively. And in the advance peak scenario, the power industry in each cluster can achieve the goal of peaking by 2030 in the 14th Five-Year Plan. The findings of this research can serve as a guide for other nations with high carbon characteristics and regional differences to transition their power industries to be low-carbon. Combined with the research results, this paper proposes the following countermeasures for the future low-carbon development of the power industry:On the Power production side. The first is to adjust the power generation structure. The large proportion of thermal power is the main problem on the power generation side, and the production of thermal power depends on coal, which is more significant in clusters 3 and 4. Therefore, controlling coal consumption is an effective way to reduce carbon emissions. We should constantly increase the proportion of clean energy power generation such as wind energy and photovoltaic to reduce the space for thermal power generation, with the proportion of thermal power generation falling to 46% in 2030 and 32% in 2040. After 2030, thermal power units should be phased out, and some of them can be converted into peak-shaving power sources or emergency backup power sources. Second, to improve the comprehensive utilization efficiency of energy. In particular, cluster 1 and cluster 4 need to strive to reduce energy consumption, accelerate the transformation of coal saving and consumption reduction through the development and introduction of new technologies to improve the energy conversion efficiency, and eliminate and shut down small and medium-sized generator sets with outdated technologies that cannot be retrofitted before 2030.The third is to develop and promote carbon capture and storage technology. Considering the technical and economic costs, in order to improve the level of decarbonization technology, it is suggested to implement it first in provinces with higher economic level in cluster 3 and cluster 4. Power transmission and distribution side. There is a need to improve the grid connection issue of clean energy power generation. Cross-provincial and cross-regional power transactions should prioritize the development of low-cost clean energy power, especially cluster 2, which is a major power export province. It is proposed to actively promote the construction of large-scale clean energy power generation transmission channels by 2030, with the goal of prioritizing clean energy delivery and constantly improving power grid mutual aid and supply guarantee capacity. It is necessary to improve the distribution network's carrying capacity to accept new energy, as well as the transmission network's intelligent grid structure, in order to promote trans-provincial transmission of new energy and increase new energy consumption capacity. It is suggested to actively to promote the construction of large-scale new energy power generation transmission channels by 2030, and increase the external supply of low-carbon energy in areas with rich clean energy resources, so as to promote the realization of optimized allocation of large-scale energy resources. Power consumption side. The first is to optimize the industrial structure. In particular, cluster 4 has a large number of industrial provinces, and it is recommended to curb the expansion of energy-intensive industries and actively develop modern service industries. Second, vigorously expand the area of replacing electric energy, promote the development of transportation electrification, tap the potential of replacing kilns and boilers in industrial production, and implement rural electrification upgrading projects to increase the proportion of electric energy in terminal consumption. Third, cultivate awareness of energy conservation. The power demand effect is the most important factor driving the increase of carbon emissions in the power industry, and cluster 4 is the most significant. Therefore, controlling power consumption through demand response and energy-saving transformation means of power demand side management is crucial to the low-carbon development of the power industry. It is recommended to implement economic measures such as reasonable industry price differentials and tiered electricity prices to reduce electricity consumption and waste, and to promote the application of energy-efficient products, equipment and technologies in high-energy-consuming industries such as steel and non-ferrous metals. Supplementary Information Below is the link to the electronic supplementary material.Supplementary file1 (DOCX 187 KB) Acknowledgements The authors would like to thank the support of the project of State Grid Hengshui Power Supply Company (Project Number kj2021-032). Authors’ contributions All authors contributed to the study conception and design. Conceptualization, supervision, writing—reviewing and editing were performed by WW. The first draft of the manuscript was written by QT. Material preparation and data collection were performed by BG. All authors read and approved the final manuscript. Availability of data and materials The dataset used and/or analyzed during the current study are available in the National Bureau of Statistics of China. http://www.stats.gov.cn/tjsj/ndsj/ Declarations Conflict of interest The authors declare no conflict of interest. Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References Chang WY Wang S Song X Zhong F Economic effects of command-and-control abatement policies under China's 2030 carbon emission goal J Environ Manage 2022 312 114925 10.1016/j.jenvman.2022.114925 35366512 Chen H Chen W Potential impacts of coal substitution policy on regional air pollutants and carbon emission reductions for China's building sector during the 13th Five-Year Plan period Energy Policy 2019 131 281 294 10.1016/j.enpol.2019.04.047 Cui X, Zhao T, Wang J (2021) Allocation of carbon emission quotas in China’s provincial power sector based on entropy method and ZSG-DEA. J Clean Prod 284. 10.1016/j.jclepro.2020.124683 De Oliveira-De JPM Effect of generation capacity factors on carbon emission intensity of electricity of Latin America & the Caribbean, a temporal IDA-LMDI analysis Renew Sustain Energy Rev 2019 101 516 526 10.1016/j.rser.2018.11.030 Demetriou E, Hadjistassou C (2021) Can China decarbonize its electricity sector? Energy Policy 148. 10.1016/j.enpol.2020.111917 Dong J, Li C, Wang Q (2021) Decomposition of carbon emission and its decoupling analysis and prediction with economic development: a case study of industrial sectors in Henan Province. J Clean Prod 321. 10.1016/j.jclepro.2021.129019 Fang K, Li C, Tang Y, He J, Song J (2022) China’s pathways to peak carbon emissions: New insights from various industrial sectors. Appl Energy 306. 10.1016/j.apenergy.2021.118039 He Y-Y, Wei Z-X, Liu G-Q, Zhou P (2020) Spatial network analysis of carbon emissions from the electricity sector in China. J Clean Prod 262. 10.1016/j.jclepro.2020.121193 He Y, Zhu L, Fan J, Li L, Liu G (2021) Life cycle assessment of CO2 emission reduction potential of carbon capture and utilization for liquid fuel and power cogeneration. Fuel Process Technol 221. 10.1016/j.fuproc.2021.106924 Hernández KD, Fajardo OA (2021) Estimation of industrial emissions in a Latin American megacity under power matrix scenarios projected to the year 2050 implementing the LEAP model. J Clean Prod 303. 10.1016/j.jclepro.2021.126921 Jandačka J Mičieta J Holubčík M Nosek R Experimental determination of bed temperatures during wood pellet combustion Energy Fuels 2017 31 3 2919 2926 10.1021/acs.energyfuels.6b03071 Jiang J-J Ye B Zhou N Zhang X-L Decoupling analysis and environmental Kuznets curve modelling of provincial-level CO2 emissions and economic growth in China: a case study J Clean Prod 2019 212 1242 1255 10.1016/j.jclepro.2018.12.116 Jiang S, Zhu Y, He G, Wang Q, Lu Y (2020) Factors influencing China’s non-residential power consumption: Estimation using the Kaya–LMDI methods. Energy 201. 10.1016/j.energy.2020.117719 Liao C, Wang S, Zhang Y, Song D, Zhang C (2019) Driving forces and clustering analysis of provincial-level CO2 emissions from the power sector in China from 2005 to 2015. J Clean Prod 240. 10.1016/j.jclepro.2019.118026 Lin D Zhang L Chen C Lin Y Wang J Qiu R Hu X Understanding driving patterns of carbon emissions from the transport sector in China: evidence from an analysis of panel models Clean Technol Environ Policy 2019 21 6 1307 1322 10.1007/s10098-019-01707-y Luo F, Guo Y, Yao M, Cai W, Wang M, Wei W (2020) Carbon emissions and driving forces of China’s power sector: Input-output model based on the disaggregated power sector. J Clean Prod 268. 10.1016/j.jclepro.2020.121925 Mardoyan A Braun P Analysis of Czech subsidies for solid biofuels Int J Green Energy 2014 12 4 405 408 10.1080/15435075.2013.841163 Maroušek J Biotechnological partition of the grass silage to streamline its complex energy utilization Int J Green Energy 2014 11 9 962 968 10.1080/15435075.2013.833930 Marousek J Gavurova B Recovering phosphorous from biogas fermentation residues indicates promising economic results Chemosphere 2022 291 Pt 1 133008 10.1016/j.chemosphere.2021.133008 34838843 Marousek J Trakal L Techno-economic analysis reveals the untapped potential of wood biochar Chemosphere 2022 291 Pt 1 133000 10.1016/j.chemosphere.2021.133000 34808200 Meng M Jing K Mander S Scenario analysis of CO2 emissions from China's electric power industry J Clean Prod 2017 142 3101 3108 10.1016/j.jclepro.2016.10.157 Qin Q, Jiao Y, Gan X, Liu Y (2020) Environmental efficiency and market segmentation: an empirical analysis of China’s thermal power industry. J Clean Prod 242. 10.1016/j.jclepro.2019.118560 Tang B Li R Yu B An R Wei Y-M How to peak carbon emissions in China's power sector: a regional perspective Energy Policy 2018 120 365 381 10.1016/j.enpol.2018.04.067 Tian Q Zhao T Yuan R An overview of the inequality in China’s carbon intensity 1997–2016: a Theil index decomposition analysis Clean Technol Environ Policy 2021 23 5 1581 1601 10.1007/s10098-021-02050-x Wang J He S Qiu Y Liu N Li Y Dong Z Investigating driving forces of aggregate carbon intensity of electricity generation in China Energy Policy 2018 113 249 257 10.1016/j.enpol.2017.11.009 Wang J Song C Yuan R CO2 emissions from electricity generation in China during 1997–2040: The roles of energy transition and thermal power generation efficiency Sci Total Environ 2021 773 145026 10.1016/j.scitotenv.2021.145026 33582330 Wang Q Jiang X-t Li R Comparative decoupling analysis of energy-related carbon emission from electric output of electricity sector in Shandong Province, China Energy 2017 127 78 88 10.1016/j.energy.2017.03.111 Wei W, Zhang P, Yao M, Xue M, Miao J, Liu B, Wang F (2020) Multi-scope electricity-related carbon emissions accounting: a case study of Shanghai. J Clean Prod 252. 10.1016/j.jclepro.2019.119789 Wen L Li Z Driving forces of national and regional CO2 emissions in China combined IPAT-E and PLS-SEM model Sci Total Environ 2019 690 237 247 10.1016/j.scitotenv.2019.06.370 31288115 Wen L Li Z Provincial-level industrial CO2 emission drivers and emission reduction strategies in China: combining two-layer LMDI method with spectral clustering Sci Total Environ 2020 700 134374 10.1016/j.scitotenv.2019.134374 31629266 Xu W Xie Y Xia D Ji L Huang G A multi-sectoral decomposition and decoupling analysis of carbon emissions in Guangdong province China J Environ Manage 2021 298 113485 10.1016/j.jenvman.2021.113485 34385114 Yan Q Wang Y Balezentis T Streimikiene D Analysis of China's regional thermal electricity generation and CO2 emissions: decomposition based on the generalized Divisia index Sci Total Environ 2019 682 737 755 10.1016/j.scitotenv.2019.05.143 31141755 Yan Q Wang Y Li Z Baležentis T Streimikiene D Coordinated development of thermal power generation in Beijing-Tianjin-Hebei region: evidence from decomposition and scenario analysis for carbon dioxide emission J Clean Prod 2019 232 1402 1417 10.1016/j.jclepro.2019.05.256 Yang J Cai W Ma M Li L Liu C Ma X Li L Chen X Driving forces of China's CO2 emissions from energy consumption based on Kaya-LMDI methods Sci Total Environ 2020 711 134569 10.1016/j.scitotenv.2019.134569 32000310 Yang X, Jia Z, Yang Z, Yuan X (2021) The effects of technological factors on carbon emissions from various sectors in China—a spatial perspective. J Clean Prod 301. 10.1016/j.jclepro.2021.126949 Yin Q, Wang Y, Xu Z, Wan K, Wang D (2022) Factors influencing green transformation efficiency in China’s mineral resource-based cities: method analysis based on IPAT-E and PLS-SEM. J Clean Prod 330. 10.1016/j.jclepro.2021.129783 Yu Y, Jin Z-x, Li J-z, Jia L (2020) Low-carbon development path research on China’s power industry based on synergistic emission reduction between CO2 and air pollutants. J Clean Prod 275. 10.1016/j.jclepro.2020.123097 Zhang C, He G, Johnston J, Zhong L (2021) Long-term transition of China’s power sector under carbon neutrality target and water withdrawal constraint. J Clean Prod 329. 10.1016/j.jclepro.2021.129765 Zhang P, Cai W, Yao M, Wang Z, Yang L, Wei W (2020) Urban carbon emissions associated with electricity consumption in Beijing and the driving factors. Appl Energy 275. 10.1016/j.apenergy.2020.115425 Zhang Y Liu C Chen L Wang X Song X Li K Energy-related CO2 emission peaking target and pathways for China's city: a case study of Baoding City J Clean Prod 2019 226 471 481 10.1016/j.jclepro.2019.04.051

0

PMC9754311

NO-CC CODE

2022-12-16 23:26:35

no

Clean Technol Environ Policy. 2022 Dec 15;:1-15

utf-8

Clean Technol Environ Policy

10.1007/s10098-022-02456-1

oa_other

==== Front Drug Saf Drug Saf Drug Safety 0114-5916 1179-1942 Springer International Publishing Cham 1265 10.1007/s40264-022-01265-1 Review Article Colchicine Drug Interaction Errors and Misunderstandings: Recommendations for Improved Evidence-Based Management http://orcid.org/0000-0002-5835-7366 Hansten Philip D. hansten@uw.edu phansten@gmail.com 1 Tan Malinda S. 2 Horn John R. 1 Gomez-Lumbreras Ainhoa 2 Villa-Zapata Lorenzo 3 Boyce Richard D. 4 Subbian Vignesh 5 Romero Andrew 6 Gephart Sheila 7 http://orcid.org/0000-0002-5006-9394 Malone Daniel C. 2 1 grid.34477.33 0000000122986657 School of Pharmacy, University of Washington, Seattle, WA USA 2 grid.223827.e 0000 0001 2193 0096 Department of Pharmacotherapy, College of Pharmacy, University of Utah, Salt Lake City, Utah, USA 3 grid.259906.1 0000 0001 2162 9738 College of Pharmacy, Mercer University, Atlanta, GA USA 4 grid.21925.3d 0000 0004 1936 9000 Department of Biomedical Informatics, School of Medicine, University of Pittsburgh, Pittsburgh, PA, USA 5 grid.134563.6 0000 0001 2168 186X College of Engineering, University of Arizona, Tucson, AZ, USA 6 grid.417332.0 0000 0000 8607 6751 Department of Pharmacy, Tucson Medical Center, Tucson, AZ USA 7 grid.134563.6 0000 0001 2168 186X College of Nursing, University of Arizona, Tucson, AZ, USA 15 12 2022 120 27 11 2022 © The Author(s), under exclusive licence to Springer Nature Switzerland AG 2022, Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. Colchicine is useful for the prevention and treatment of gout and a variety of other disorders. It is a substrate for CYP3A4 and P-glycoprotein (P-gp), and concomitant administration with CYP3A4/P-gp inhibitors can cause life-threatening drug–drug interactions (DDIs) such as pancytopenia, multiorgan failure, and cardiac arrhythmias. Colchicine can also cause myotoxicity, and coadministration with other myotoxic drugs may increase the risk of myopathy and rhabdomyolysis. Many sources of DDI information including journal publications, product labels, and online sources have errors or misleading statements regarding which drugs interact with colchicine, as well as suboptimal recommendations for managing the DDIs to minimize patient harm. Furthermore, assessment of the clinical importance of specific colchicine DDIs can vary dramatically from one source to another. In this paper we provide an evidence-based evaluation of which drugs can be expected to interact with colchicine, and which drugs have been stated to interact with colchicine but are unlikely to do so. Based on these evaluations we suggest management options for reducing the risk of potentially severe adverse outcomes from colchicine DDIs. The common recommendation to reduce the dose of colchicine when given with CYP3A4/P-gp inhibitors is likely to result in colchicine toxicity in some patients and therapeutic failure in others. A comprehensive evaluation of the almost 100 reported cases of colchicine DDIs is included in table form in the electronic supplementary material. Colchicine is a valuable drug, but improvements in the information about colchicine DDIs are needed in order to minimize the risk of serious adverse outcomes. Supplementary Information The online version contains supplementary material available at 10.1007/s40264-022-01265-1. http://dx.doi.org/10.13039/100000133 Agency for Healthcare Research and Quality R01HS025984 Malone Daniel C. ==== Body pmcKey Points Colchicine is a valuable drug for many indications, but it has a number of potentially serious drug interactions. Most current sources of colchicine drug interaction information have errors of omission or commission regarding which drugs interact clinically with colchicine. Adverse colchicine drug interactions can almost always be prevented through proper management and patient education. Introduction Colchicine has been a valuable drug in the prevention and treatment of gout for many centuries, and it is also approved for use in familial Mediterranean fever. Colchicine has a variety of off-label uses, including pericarditis, various dermatologic disorders, Behçet’s disease, and more recently it has been used to reduce cardiovascular events in patients with coronary artery disease [1–4]. Colchicine has also been tested in patients with COVID-19 in numerous clinical trials, but the results to date have been disappointing [5, 6]. Colchicine is still under investigation for treatment of long COVID [7], and it can be used to treat the pericarditis that can occur in COVID patients [8]. As the indications for colchicine increase beyond gout, it is logical to expect to see more adverse drug–drug interactions (DDIs) because more people will receive colchicine, and the prescribers may have little previous experience using the drug. In clinical trials of low-dose colchicine (0.5 mg/day) following myocardial infarction, few DDIs have been observed [2]. This lack of DDIs in low-dose colchicine trials is consistent with the case reports detailed in the electronic supplementary material (ESM), in which the vast majority of adverse colchicine DDIs involved colchicine doses > 0.5 mg. Nonetheless, it is likely that the risk of colchicine DDIs is lower in these clinical trials as opposed to the general clinical setting in which patients may receive higher colchicine doses and where patients are not as carefully screened for risk factors, concomitant therapy, and may not be carefully monitored for adverse outcomes [9]. Given that colchicine DDIs can result in serious or fatal outcomes, it is important that health care professionals (i) know which drugs can interact with colchicine, and (ii) understand how to manage such DDIs to minimize the risk of patient harm. The purpose of this paper is to provide that information, based on the evidence provided in the literature and the general principles of drug interaction management. Colchicine Toxicity Colchicine has a narrow therapeutic index and must be dosed carefully to minimize the risk of toxicity. Toxicity tends to occur when the serum colchicine concentration exceeds 3.0 μg/L [10]. While colchicine has been used safely in many patients for many centuries, colchicine toxicity can occur when patients develop excessive serum concentrations, often resulting from some combination of excessive dosing, drug interactions, and impaired renal or hepatic function [11]. Once serious colchicine toxicity occurs it can be very difficult to treat, especially if the toxicity damages the kidneys and liver, which further impairs the ability of the person to eliminate colchicine. Even when the patient survives, elevated colchicine concentrations may persist for weeks after it is discontinued [12–14] resulting in continuing damage to tissues and organs. Colchicine is not dialysable, so dialysis does not shorten the process. Indeed, patients on maintenance hemodialysis appear to be at greater risk from colchicine DDIs [15–17]. Colchicine toxicity usually manifests as gastrointestinal, musculoskeletal, hematologic, and—with severe toxicity—cardiovascular. These are summarized in Table 1. Over 100 DDI cases of colchicine-related toxicity, many of them life-threatening or fatal, have been published in case reports, case series, and adverse reaction reporting systems [18, 19]. A detailed analysis of colchicine DDI cases published in the medical literature is provided in the online Supplemental Table (see ESM).Table 1 Clinical findings of colchicine toxicity Severity Signs and symptoms Mild Gastrointestinal: diarrhea, nausea, vomiting, abdominal pain Miscellaneous: fatigue, lethargy, malaise, insomnia Moderate Neuromuscular: muscle pain, muscle weakness, paresthesias Hematologic: moderate neutropenia and/or moderate thrombocytopenia Respiratory: shortness of breath, cough Dermatologic: alopecia (usually late) Severe/fatal Neuromuscular: rhabdomyolysis, atonia, dark brown urine Hematologic: pancytopenia (infections, fever, bleeding) Multiorgan failure: renal failure, liver failure Cardiovascular: cardiac arrhythmias, cardiac failure, hypotension, cardiac arrest Colchicine Pharmacokinetics Despite colchicine being used for many centuries, its pharmacokinetics have only been studied in the past few decades after suitable assay techniques were developed. Research is still being conducted, particularly on the effect of transporters on colchicine disposition. Absorption Colchicine is a lipid-soluble alkaloid and undergoes extensive first-pass metabolism resulting in an absolute bioavailability of approximately 30–50% [11, 20, 21]. Bioavailability can vary substantially from one patient to another, probably due to the numerous enzymes and transporters involved during absorption. In one study of healthy subjects, mean bioavailability was 45%, but it varied among subjects from 28 to 88% [22]. Colchicine is a substrate for CYP3A4 and P-gp in the intestine, and P-gp appears to transport colchicine back into the intestinal lumen. It has been proposed that colchicine may be a substrate for transporters such as MRP2 and OATP2B1 [23, 24] but their role remains to be established. Distribution Colchicine is rapidly distributed, and the apparent volume of distribution of colchicine is large; it is usually about 5–10 L/kg but can range from 2 to 12 L/kg. The wide distribution and low (nanogram per milliliter) concentrations in the plasma are consistent with the failure of dialysis to effectively treat colchicine toxicity. Binding to serum protein is about 40%, again with high variability. Colchicine crosses the placenta and is distributed into breast milk [11, 21]. There are no reported DDIs based on changes in colchicine distribution. Metabolism Colchicine metabolism takes place primarily through demethylation by CYP3A4 in the intestine and liver. Other CYP isozymes and phase II metabolism do not appear to play a significant role in colchicine metabolism. Elimination The kidneys are important in the elimination of colchicine, but estimates of the percentage of total colchicine clearance by the kidneys have varied [21]. In 12 healthy subjects given colchicine 1 mg, 40–65% was recovered unchanged in the urine [25], but others have found lower percentages for renal elimination [21]. Biliary excretion also appears to be an important pathway of colchicine elimination, and P-gp appears to be involved in both the renal tubular secretion and the biliary excretion of colchicine [21, 26]. The half-life in healthy subjects is generally about 15–30 h [21, 22]. In one small study, total body clearance was about twice as high and area under the concentration–time curve (AUC) was about four times smaller for healthy subjects compared with elderly patients [22]. Mechanisms of Colchicine Drug Interactions Colchicine can interact with other drugs by a variety of different mechanisms (see Table 2). For some colchicine DDIs, more than one mechanism is involved. For example, several drugs that cause myopathy also inhibit CYP3A4 and/or P-gp (e.g., amiodarone, cyclosporine, statins). Colchicine is rarely the perpetrator in DDIs with other drugs, and in this paper we address only colchicine as the victim drug in DDIs.Table 2 Sites of potential interaction with colchicine Process Mediated by Drugs that may interact Absorption CYP3A4 and P-gp in wall of small intestinea Inhibitors or inducers of CYP3A4 and P-gp in wall of small intestine Hepatic metabolism CYP3A4 and P-gp in liver Inhibitors or inducers of CYP3A4 and P-gp in liver Hepatotoxic drugs (theoretically) Biliary excretion P-gp in liver Inhibitors or inducers of P-gp in liver Renal excretion Glomerular filtration P-gp in kidney Nephrotoxic drugs (theoretically) Inhibitors or inducers of P-gp in kidney Myotoxicity Possible additive effect with myotoxic effect of colchicine Drugs that cause myopathy aColchicine undergoes enterohepatic circulation, it may be exposed to CYP3A4 and P-gp more than once Inhibition of CYP3A4 and/or P-gp Colchicine is a substrate for CYP3A4 in the intestine and liver, and undergoes considerable first-pass metabolism resulting in bioavailability of 30–50% [11, 21]. It is also a substrate for P-gp in the small intestine, liver, and kidneys [11, 23]. Colchicine undergoes enterohepatic circulation as parent drug and metabolites. Many drugs that produce combined CYP3A4/P-gp inhibition have been shown to increase colchicine serum concentrations and cause colchicine toxicity as we describe elsewhere in this paper. Because it is well documented that combined CYP3A4/P-gp inhibitors increase colchicine AUC, drugs that inhibit either CYP3A4 or P-gp alone are also often assumed to be capable of increasing colchicine AUC. The vast majority of drugs that inhibit CYP3A4 also inhibit P-gp, so this is not an issue for most colchicine DDIs. It is not established, however, that inhibition of either CYP3A4 or P-gp alone is enough to cause clinically significant DDIs with colchicine. In fact, the available evidence, although limited, suggests otherwise. Drugs that primarily inhibit one or the other—such as voriconazole (CYP3A4 inhibitor) or propafenone (P-gp inhibitor)—have not had much effect on colchicine pharmacokinetics. Voriconazole is classified by the US FDA as a ‘strong’ CYP3A4 inhibitor, and it can substantially increase the AUC of CYP3A4 substrates [27, 28]. Voriconazole may not inhibit P-gp, however, suggested by a lack of voriconazole on digoxin pharmacokinetics in healthy subjects [29]. A pharmaceutical company applying to the US Food and Drug Administration (FDA) for approval of their colchicine product performed a non-randomized, open-label, crossover study in 12 healthy subjects given colchicine 0.6 mg before and after voriconazole 200 mg twice daily for 5 days. Voriconazole did not affect colchicine pharmacokinetics [30]. On the other hand, propafenone inhibits P-gp but is not known to inhibit CYP3A4. A pharmaceutical company applying to the US FDA for approval of their colchicine product conducted a non-randomized, open label, crossover study in nine healthy male subjects who were given a single dose of colchicine 0.6 mg before and after propafenone 225 mg twice daily for 5 days. Propafenone did not appear to affect colchicine pharmacokinetics [30]. However, one cannot rule out that a larger daily dose of propafenone given for a longer period might interact with colchicine. It is also possible that P-gp inhibition without CYP3A4 inhibition has little effect on colchicine pharmacokinetics. A tentative conclusion from the studies of voriconazole and propafenone is that inhibition of both CYP3A4 and P-gp is needed in order to manifest clinically important effects on colchicine pharmacokinetics. Nonetheless, one cannot rule out that inhibition of either CYP3A4 or P-gp alone may be sufficient to interact with colchicine, especially in patients predisposed to colchicine toxicity such as those with renal or hepatic toxicity, or those with reduced activity of P-gp or CYP3A4 due to genetics. Enzyme Inducers Enzyme inducers such as rifampin, barbiturates, carbamazepine, efavirenz, lumacaftor, phenytoin, primidone, and St. John’s wort increase CYP3A4/P-gp activity, and would be expected to reduce colchicine AUC. Case reports suggest that both rifampin and carbamazepine can substantially reduce colchicine concentrations [31, 32]. (See case report details in the online Supplemental Table in the ESM). Lesinurad appears to be a modest CYP3A4 inducer, and it produced modest reductions in colchicine AUC in healthy subjects [33]. Although the evidence for the effect of CYP3A43 inducers on colchicine is limited, it seems very likely that patients on enzyme inducers are at increased risk of subtherapeutic colchicine concentrations. Additive Myotoxic Effects Colchicine therapy alone has been associated with myopathy and rhabdomyolysis [34–40], and it is possible that additive myotoxic effects with other drugs can occur. Myotoxicity has been reported in numerous patients receiving colchicine along with drugs that can independently lead to myotoxicity such as certain statins and cyclosporine. (See case report details in the online Supplemental Table, ESM.) Although it is thought that these reactions are due to additive myotoxic effects, it cannot yet be ruled out that pharmacokinetic DDIs may also occur, involving CYP3A4, P-gp, OATP [24] or MRP2 [23, 41]. Evidence to date suggests that not all statins pose the same risk when combined with colchicine, as discussed in Table 3. The signs and symptoms of myotoxicity in these cases often included muscle weakness and/or muscle pain, and sometimes darkened urine; myopathy is also often accompanied by nonspecific symptoms such as fatigue and malaise, and shortness of breath.Table 3 Management options for colchicine drug interactions Drugs Evidence for drug interaction (DDI) ORCA classa and management options Management if concurrent use needed Antiarrhythmics  Amiodarone Amiodarone inhibits CYP3A4/P-gp and would be expected to ↑ colchicine AUC; [61] several case reports suggest that a DDI may occurb ORCA Class 2: Avoid if possiblec Use alternative: other than amiodarone and dronedarone, most other antiarrhythmics are not known to inhibit CYP3A4 and P-gp Or consider alternative to colchicined Reduce colchicine dose 50–75% Monitor for colchicine toxicity Advise patient about colchicine toxicitye  Dronedarone Dronedarone inhibits CYP3A4/P-gp and theoretically would be expected to ↑ colchicine AUC [62, 63] ORCA Class 2: Avoid if possiblec Use alternative: other than amiodarone and dronedarone, most other antiarrhythmics are not known to inhibit CYP3A4 and P-gp Or consider alternative to colchicined Reduce colchicine dose 50–75% Monitor for colchicine toxicity Advise patient about colchicine toxicitye  Propafenone Propafenone inhibits P-gp but has little effect on CYP3A4; no colchicine DDI was found in healthy subjects with propafenone 450 mg/d × 5 days [30, 56] ORCA Class 4: Low risk Concurrent use need not be avoided, but some patients might have increased colchicine levels especially if they have low CYP3A4 activity due to other drugs or genetics Normal monitoring for colchicine toxicity  Quinidine Quinidine is a potent inhibitor of P-gp, but not CYP3A4 [64]; it is not known if P-gp inhibition alone affects colchicine pharmacokinetics, no clinical data are available ORCA Class 3: Assess risk; take action if needed Concurrent use need not be avoided, but increased colchicine levels possible, especially in predisposed patients (e.g., on CYP3A4 inhibitors, severe renal or hepatic impairment)c Monitor for altered colchicine effect if quinidine is started, stopped or changed in dosage Advise patient about colchicine toxicitye  Ranolazine Ranolazine appears to be a modest inhibitor of CYP3A4 and P-gp [65–67]; it is not known if inhibition of CYP3A4 and P-gp is large enough for interaction ORCA Class 3: Assess risk; take action if needed Concurrent use need not be avoided, but increased colchicine levels possible, especially in predisposed patients (e.g., on CYP3A4 inhibitors, severe renal or hepatic impairment)c Monitor for altered colchicine effect if ranolazine is started, stopped or changed in dosage Advise patient about colchicine toxicitye Azole antifungals  Fluconazole Fluconazole is a dose-dependent inhibitor of CYP3A4 that probably has little effect on P-gp [27, 68–70]; fluconazole 400 mg on day 1 and 200 mg on days 2–5 ↑ colchicine AUC by 40% in healthy subjects [56]; one case of colchicine toxicity with fluconazole in patient with renal failureb ORCA Class 3: Assess risk; take action if needed Concurrent use need not be avoided, but increased risk with renal impairmentc,f or if patient has low P-gp activityg Or consider alternative to fluconazole: voriconazole may be less likely to interact with colchicine (see below); terbinafine does not inhibit CYP3A4 If no renal impairment or ↓ P-pg activity: Monitor for colchicine toxicity particularly during start of fluconazole therapy Advise patient about colchicine toxicitye With renal impairment or ↓ P-gp activity: Reduce colchicine dose 50–75% Monitor for colchicine toxicity Advise patient about colchicine toxicitye  Itraconazole Itraconazole inhibits CYP3A4/P-gp [27] and would be expected to ↑ colchicine AUC as does ketoconazole (see below) ORCA Class 2: Avoid if possiblec Consider stopping colchicine during short-term itraconazole Or use alternative. Fluconazole (see above) and voriconazole (see below) are less likely to interact; terbinafine does not inhibit CYP3A4 Or consider alternative to colchicined Reduce colchicine dose 50–75% Monitor for colchicine toxicity Advise patient about colchicine toxicitye  Ketoconazole Ketoconazole inhibits CYP3A4/P-gp. A study in 24 healthy subjects found a mean 212% ↑ in colchicine AUC after ketoconazole 200 mg BID for 5 days; one subject had a 420% ↑ in AUC [25] ORCA Class 2: Avoid if possiblec Consider stopping colchicine during short-term ketoconazole Or use alternative. Fluconazole (see above) and voriconazole (see below) are less likely to interact; terbinafine does not inhibit CYP3A4 Or consider alternative to colchicined Reduce colchicine dose 50–75% Monitor for colchicine toxicity Advise patient about colchicine toxicitye  Voriconazole Voriconazole is a strong CYP3A4 inhibitor, but appears to have little effect on P-gp [27–29]; a study in healthy subjects found no effect of voriconazole 200 mg BID for 5 days on colchicine AUC [30] ORCA Class 4: Low risk Concurrent use need not be avoided, but some patients might have increased colchicine levels, especially if P-gp activity is lowg Normal monitoring for colchicine toxicity Antivirals—protease inhibitors and adjuvants  Atazanavir  Cobicistat  Darunavirh  Fosamprenavir  Indinavir  Nelfinavir  Ritonavir  Saquinavir  Telaprevir These agents inhibit CYP3A4 and P-gp and would be expected to ↑ colchicine AUC; a study in 18 healthy subjects found a mean 296% ↑ in colchicine AUC after ritonavir 100 mg/day for 5 days; one subject had a 924% ↑ in colchicine AUCi [25, 45] ORCA Class 2: Avoid if possiblec Use alternative: For most patients it may be easier to avoid colchicine than the antiviral agentd. Stop colchicine before antiviral is started, and for 3–4 days after antiviral is stopped. If concurrent use is deemed necessary (which should very rarely be the case), see column to the right. Ritonavir + nirmatrelvir (Paxlovid) is normally given for 5 days, so stopping colchicine during Paxlovid therapy would usually be the best management Reduce colchicine dose 50–75% Monitor for colchicine toxicity Advise patient about colchicine toxicitye  Tipranavir Unlike many other protease inhibitors, tipranavir tends to ↑ P-gp activity, which might reduce colchicine AUCj [71] ORCA Class 3: Assess risk; take action if needed Concurrent use need not be avoided, but some patients may have reduced colchicine concentrations; more data are needed Monitor for altered colchicine effect if tipranavir is started, stopped, or changed in dosage; adjust colchicine dose as needed; note that the onset and offset of enzyme induction may occur slowly over 1–2 weeks Antivirals—miscellaneous  Boceprevir Boceprevir inhibits CYP3A4 [72], but its effect on P-pg is not known. Effect on colchicine requires further study ORCA Class 3: Assess risk; take action if needed Concurrent use need not be avoided, but some patients might have increased colchicine levels, especially if P-gp activity is lowg Monitor for altered colchicine effect if boceprevir is started, stopped or changed in dosage Advise patient about colchicine toxicitye  Letermovir Letermovir inhibits CYP3A4 [73], but may not affect P-pg [74]. Effect on colchicine requires further study ORCA Class 3: Assess risk; take action if needed Concurrent use need not be avoided, but some patients might have increased colchicine levels, especially if P-gp activity is lowg Monitor for altered colchicine effect if letermovir is started, stopped or changed in dosage Advise patient about colchicine toxicitye  Efavirenz  Etravirine  Nevirapine These enzyme inducers are known to induce CYP3A4, and are likely to reduce colchicine serum concentrations ORCA Class 3: Assess risk; take action if needed Concurrent use need not be avoided, but some patients may have reduced colchicine concentrations; more data are needed Monitor for altered colchicine effect if these agents are started, stopped, or changed in dosage; adjust colchicine dose as needed; note that the onset and offset of enzyme induction may occur slowly over 1–2 weeks Calcineurin inhibitors  Cyclosporine Cyclosporine inhibits CYP3A4 and P-gp; a study in 23 healthy subjects found a mean 259% ↑ in colchicine AUC after a single 100-mg dose of cyclosporine; one subject had a 512% ↑ [25, 45, 75]; possible additive myopathyk; over 30 cases of colchicine–cyclosporine DDI reportedb ORCA Class 2: Avoid if possiblec The risk would almost always outweigh the benefit Use alternative: If appropriate, tacrolimus appears less likely to interact with colchicine (see below); many other immunosuppressants are not known to inhibit CYP3A4 and P-gp Or consider alternative to colchicined Reduce colchicine dose 50–75% Monitor for colchicine toxicity, especially for signs of myopathy Advise patient about colchicine toxicitye  Tacrolimus Tacrolimus inhibits P-gp somewhat, but may be only weak inhibitor of CYP3A4; Two small studies found increased colchicine AUC with tacrolimus, but less than with cyclosporine [16, 75] ORCA Class 3. Assess risk; take action if needed Colchicine levels are likely to increase somewhat, especially in predisposed patients (e.g., on CYP3A4 inhibitors, or with severe renal or hepatic impairment)c Consider alternative to tacrolimus or colchicined Monitor for altered colchicine effect if tacrolimus started, stopped, or changed in dosage Advise patient about colchicine toxicitye Calcium-channel blockers (CCBs)  Diltiazem Diltiazem inhibits CYP3A4 and P-gp; in 20 healthy subjects, diltiazem 240 mg BID for 7 days ↑ colchicine AUC by 93%; one subject had a 339% increase [25, 45]; a patient on diltiazem developed fatal colchicine toxicity, but causality was not establishedb ORCA Class 2: Avoid if possiblec Use alternative. If appropriate, use an alternative to diltiazem or verapamil; most other CCBs such as dihydropyridines appear to have little effect on CYP3A4 or P-gp Or consider alternative to colchicined Reduce colchicine dose 50–75% Monitor for colchicine toxicity, especially for signs of myopathy Advise patient about colchicine toxicitye  Verapamil Verapamil inhibits CYP3A4 and P-gp; in 24 healthy subjects, verapamil 240 mg BID for 5 days ↑ colchicine AUC by 103%; one subject had a 217% increase [25, 45]; a patient on verapamil developed serious colchicine toxicityb ORCA Class 2: Avoid if possiblec Use alternative. If appropriate, use an alternative to verapamil or diltiazem; most other CCBs such as dihydropyridines appear to have little effect on CYP3A4 or P-gp Or consider alternative to colchicined Reduce colchicine dose 50–75% Monitor for colchicine toxicity, especially for signs of myopathy Advise patient about colchicine toxicitye Enzyme inducers  Carbamazepine (CBZ) Carbamazepine induces CYP3A4 and P-gp, and would be expected to reduce colchicine AUC; case report of very low colchicine levels in patient on CBZb ORCA Class 2: Avoid if possible It may be difficult to achieve therapeutic colchicine levels in presence of CBZ Use alternative. If appropriate, use an alternative to the CBZ or colchicined Monitor for altered colchicine effect if CBZ is started, stopped, or changed in dosage; adjust colchicine dose as needed; note that the onset and offset of enzyme induction may occur slowly over 1–2 weeks  Rifampin Rifampin induces CYP3A4 and P-gp, and would be expected to reduce colchicine AUC; case report of low colchicine levels in patient on rifampinb ORCA Class 2: Avoid if possible It may be difficult to achieve therapeutic colchicine levels in presence of rifampin Use alternative. If appropriate, use an alternative to rifampin or colchicined Monitor for altered colchicine effect if CBZ is started, stopped, or changed in dosage; adjust colchicine dose as needed; note that the onset and offset of enzyme induction may occur slowly over 1–2 weeks  Barbiturates  Bosentan  Dabrafenib  Lumacaftor  Phenytoin  Primidone  Rifabutin  Rifapentine  St. John’s wort  Topiramate These enzyme inducers are known to induce CYP3A4, and are likely to reduce colchicine serum concentrations ORCA Class 2: Avoid if possible It may be difficult to achieve therapeutic colchicine levels in presence of these enzyme inducers Use alternative. If appropriate, use an alternative to the enzyme inducer or colchicined Monitor for altered colchicine effect if inducer is started, stopped, or changed in dosage; adjust colchicine dose as needed; note that the onset and offset of enzyme induction may occur slowly over 1–2 weeks Foods  Grapefruit Grapefruit inhibits CYP3A4, but appears to be a weak P-gp inhibitor; a single case of life-threatening colchicine toxicity was reported in an 8-year-old girlb and in vitro evidence [76] suggest a DDI; but study in 21 healthy subjects given 240 mL grapefruit juice BID for 4 days found no effect on colchicine AUCl [24] ORCA Class 3. Assess risk; take action if needed Although evidence for an interaction is weak, it might occur in patients with low P-gp activityg It would be prudent to advise patents to drink juices other than grapefruit If the patient drinks grapefruit juice, advise them to avoid ingesting more than 240 mL/day  Seville oranges A study in 23 healthy subjects given 240 mL Seville orange juice BID for 4 days found a small decrease in colchicine AUC [24] ORCA Class 4. Low risk Concurrent use need not be avoided, but a slight reduction in colchicine AUC may occur Normal monitoring for colchicine toxicity HMG-CoA reductase inhibitors  Atorvastatin A study in 24 healthy subjects found a mean 24% ↑ in colchicine AUC after atorvastatin 40 mg/day for 14 days [77]; due to weak P-gp inhibition by atorvastatin? 7 case reports of myopathy with combinationb; additive myotoxicity? [58, 59] ORCA Class 3. Assess risk; take action if needed Concurrent use need not be avoided, especially if the colchicine dose is 0.6 mg/day or less If colchicine dose is > 0.6 mg/day, myopathy risk may be increased in predisposed patients, (i.e. impaired renal function, elevated statin levels due to dose, DDIs, etc.)c,m Consider alternative: fluvastatin, pravastatin, and rosuvastatin may be less likely to interact Monitor for colchicine toxicity, especially for signs of myopathy Advise patient to report evidence of myopathy (muscle weakness, myalgia, dark urine)  Fluvastatin Not expected to affect colchicine AUC; isolated cases of rhabdomyolysis reported with combination, but causality was not establishedb; additive myotoxicity? [58, 59] ORCA Class 4: Low risk Concurrent use need not be avoided; it is possible that the myopathy risk is increased in predisposed patients, but more data are neededn Normal monitoring for colchicine toxicity, with emphasis on signs of myopathy (muscle weakness, myalgia, dark urine)  Lovastatin Possible P-gp inhibitory effect of lovastatin? [78] Possible additive myotoxicity? [58, 59]. Isolated cases of myopathyb; lovastatin has similar pharmacokinetic properties as simvastatin (see below) ORCA Class 3: Assess risk; take action if needed Concurrent use need not be avoided, especially if the colchicine dose is 0.6 mg/day or less If colchicine dose is > 0.6 mg/day, myopathy risk may be increased with renal impairment or elevated statin levels due to dose, DDIs, etc.)c,m Consider alternative: fluvastatin, pravastatin, and rosuvastatin may be less likely to interact Monitor for colchicine toxicity, especially for signs of myopathy Advise patient to report evidence of myopathy (muscle weakness, myalgia, dark urine)  Pravastatin Not expected to affect colchicine AUC; isolated cases of myopathy reported with combination, but causality was not establishedb; additive myotoxicity? [58, 59] ORCA Class 4: Low risk Concurrent use need not be avoided; it is possible that the myopathy risk is increased in predisposed patients, but more data are neededn Normal monitoring for colchicine toxicity, with emphasis on signs of myopathy (muscle weakness, myalgia, dark urine)  Rosuvastatin Not expected to affect colchicine AUC; isolated cases of myopathy reported, but causality was not establishedb; additive myotoxicity? [58, 59] Combination has been used to treat COVID [79] ORCA Class 4: Low risk Concurrent use need not be avoided; it is possible that the myopathy risk is increased in predisposed patients, but more data are neededn Normal monitoring for colchicine toxicity, with emphasis on signs of myopathy (muscle weakness, myalgia, dark urine)  Simvastatin Possible P-gp inhibitory effect of simvastatin? [80] Numerous case reports of myopathy reported with combination, usually in patients with preexisting renal impairmentb; possible additive myotoxicity [58, 59] ORCA Class 3: Assess risk; take action if needed Concurrent use need not be avoided, especially if the colchicine dose is 0.6 mg/day or less If colchicine dose is > 0.6 mg/day, myopathy risk may be increased with renal impairment or elevated statin levels due to dose, DDIs, etc.c,m Consider alternative: fluvastatin , pravastatin, and rosuvastatin may be less likely to interact Monitor for colchicine toxicity, especially for signs of myopathy Advise patient to report evidence of myopathy (muscle weakness, myalgia, dark urine) Macrolide antibiotics  Azithromycin A study in 21 healthy subjects found a mean 57% ↑ in colchicine AUC after azithromycin 500 mg on day 1, then 250 mg/day for 4 days; one subject had a 241% increase [25, 45]; azithromycin is a modest P-gp inhibitor ORCA Class 3. Assess risk; take action if needed Colchicine levels are likely to increase somewhat, especially in predisposed patients (e.g., on CYP3A4 inhibitors, or with severe renal or hepatic impairment)c Consider alternative to antibiotic or colchicined Monitor for altered colchicine effect if azithromycin is started, stopped or changed in dosage Advise patient about colchicine toxicitye  Clarithromycin A study in 23 healthy subjects found a mean 282% ↑ in colchicine AUC after clarithromycin 500 mg/day for 7 days; one subject had a 852% ↑; clarithromycin inhibits both CYP3A4 and P-gp; fatalities reported in case series [19], among >2 dozen case reports,b and ADR reporting systems [18, 81] ORCA Class 1. Avoid combination The risk would almost always outweigh the benefit Consider stopping colchicine during clarithromycin therapy Or use alternative to clarithromycin; azithromycin has a smaller effect on colchicine (see above); many other antibiotics have little effect on CYP3A4 or P-gp It should not be necessary to use colchicine and clarithromycin together. If it is absolutely necessary: Reduce colchicine dose by 75% Monitor for colchicine toxicity Advise patient about colchicine toxicitye  Erythromycin Erythromycin inhibits CYP3A4 but may inhibit P-gp less than clarithromycin; isolated DDI case reportsb ORCA Class 2: Avoid if possiblec Consider stopping colchicine during erythromycin therapy Or use alternative. Consider an alternative to erythromycin; azithromycin has a smaller effect on colchicine (see above); many other antibiotics have little effect on CYP3A4 or P-gp Reduce colchicine dose 50–75% Monitor for colchicine toxicity, especially for signs of myopathy Advise patient about colchicine toxicitye Miscellaneous CYP3A4 inhibitors  Aprepitant  Ceritinib  Conivaptan  Crizotinib  Fedratinib  Idelalisib  Imatinib  Lapatinib  Lomitapide  Lefamulin  Mifapristone  Nefazodone  Quinupristin  Ribociclib  Telithromycin All of these drugs inhibit CYP3A4 and many are also known to inhibit P-gp; no information is available on the effect of these drugs on colchicine AUC, but interactions should be considered possible until proven otherwise ORCA Class 2: Avoid if possiblec Use alternative. If appropriate, us an alternative to the CYP3A4 inhibitor that does not inhibit CYP3A4 Or consider alternative to colchicined Reduce colchicine dose 50–75% Monitor for altered colchicine response if CYP3A4 inhibitor started, stopped, or changed in dosage Advise patient about colchicine toxicityd Quinolines  Chloroquine Chloroquine alone can cause myopathy [82], so additive myotoxicity with colchicine is possible; no supporting clinical data ORCA Class 4. Low risk Concurrent use need not be avoided; it is possible that the myopathy risk is increased in predisposed patients, but more data are needed Normal monitoring for colchicine toxicity, with emphasis on signs of myopathy  Hydroxychloroquine (HQ) HQ alone can cause myopathy; additive myotoxicity possible and single case of myopathy reportedb; HQ might inhibit P-gp, but no PK studies with colchicine; HQ has been used with colchicine to treat various disorders [83–85]   ORCA Class 4. Low risk Concurrent use need not be avoided; it is possible that the myopathy risk is increased in predisposed patients, but more data are needed Normal monitoring for colchicine toxicity, with emphasis on signs of myopathy Other drugs  Digoxin Digoxin is purported to cause myopathy when combined with colchicine [25, 55] but supporting evidence is lacking; the fact that digoxin is a P-gp substrate is not proof that it inhibits P-gp ORCA Class 4. Low risk Concurrent use need not be avoided Normal monitoring for colchicine toxicity  Disulfiram Evidence that disulfiram inhibits P-gp or CYP3A4 is weak [86, 87]; single case report of DDI with colchicine, but causal relationship doubtfulb ORCA Class 4. Low risk Concurrent use need not be avoided Normal monitoring for colchicine toxicity  Fibrates Fenofibrate and gemfibrozil known to cause myopathy, with or without statins [88]; additive myotoxic effect with colchicine? Evidence that fenofibrate inhibits P-gp is weak [89]; one case of myopathy with gemfibrozil + colchicineb ORCA Class 3: Assess risk; take action if needed Concurrent use need not be avoided; it is possible that the myopathy risk is increased in predisposed patients, but more data are needed. Risk may be increased if patient is also taking a statin Monitor for colchicine toxicity, especially for signs of myopathy Advise patient to report evidence of myopathy (muscle weakness, myalgia, dark urine)  Lesinurad Lesinurad may be a modest CYP3A4 inducer [90]; PK study in healthy subjects found modest ↓ in colchicine AUC, but lesinurad dose was high (400–600 mg/day) [33] ORCA Class 3: Assess risk; take action if needed Concurrent use need not be avoided, but some patients may have reduced colchicine concentrations; more data are needed Monitor for altered colchicine effect if lesinurad is started, stopped, or changed in dosage. Adjust colchicine dose as needed  Nivolumab Thrombocytopenia reported in a patient on nivolumab and colchicineb; nivolumab may inhibit CYP3A4 by upregulating cytokines [91] but an effect on colchicine is speculative ORCA Class 3: Assess risk; take action if needed This DDI is largely theoretical, and concurrent use need not be avoided, some patients might have increased colchicine AUC, but more data are needed to establish a causal effect Monitor for altered colchicine effect if nivolumab is started, stopped, or changed in dosage. Adjust colchicine dose as needed  Sunitinib Colchicine toxicity reported in a patient on sunitinibb, but the colchicine toxicity was more likely due to the excessive dose of colchicine, and the effect of concurrent diltiazem (see above) ORCA Class 4: Low risk Concurrent use need not be avoided Normal monitoring for colchicine toxicity ADRs adverse drug reactions, AUC area under the concentration–time curve, BID twice daily, DDIs drug–drug interactions, P-gp P-glycoprotein, PK pharmacokinetics, ↑ indicates increase, ↓ indicates decrease aORCA: OpeRational ClassificAtion of drug interactions [92]: Class 1 = Avoid combination; Class 2 = Avoid combination unless benefit outweighs risk; Class 3 = Assess risk and take action if needed; Class 4 = Low risk: no special precautions; Class 5 = No interaction bCase report details, including evaluation of causality, are presented in Online Resource (see ESM) cDrug interactions that result in elevated colchicine plasma concentrations tend to be more dangerous in patients with renal impairment. Consider patients with significant renal impairment (e.g., creatinine clearance [CrCl] < 60 mL/min) to be at somewhat higher risk, and patients with severe renal disease (e.g., CrCl < 30 mL/min) at much higher risk of colchicine toxicity from drug interactions. For all drug interactions listed as ‘Class 2’, consider them contraindicated in patients with severe renal disease. The effect of liver disease is more theoretical, but assume ORCA class 2 DDIs are contraindicated if patient has a Child-Pugh score of C dIf colchicine is being used for acute gout, consider using alternative gout therapy. If colchicine is being used chronically for gout prophylaxis, consider the risk of colchicine toxicity vs the expected benefit of gout prevention. If colchicine is being used for familial Mediterranean fever, cardiovascular disease, dermatologic disorders, Covid-19, or other disorders, it may be more difficult to find alternatives for colchicine. For some disorders such as pericarditis, colchicine may be deemed necessary ePatient should report colchicine toxicity such as severe diarrhea, protracted vomiting, muscle weakness, muscle pain, dark urine, fever or other signs of infection, unusual bleeding or bruising, severe fatigue fFluconazole and colchicine are both eliminated renally, so patients with severe renal impairment will have increased serum concentrations of both drugs. And since the ability of fluconazole to inhibit CYP3A4 is related to serum concentrations, there will be increased CYP3A4 inhibition gBased on available evidence, drugs that inhibit CYP3A4 but not P-gp have little effect on colchicine AUC. But the risk of a DDI with colchicine would theoretically be increased if CYP3A4 inhibitors are given to patients with low P-gp activity due to other drugs or genetics hDarunavir inhibits CYP3A4, but may induce P-gp somewhat. Theoretically, this might mitigate any increase in colchicine serum concentrations iLong-term ritonavir therapy can result in enzyme induction [93], so it is not clear if these results apply to patients on chronic ritonavir jTipranavir is generally given with ritonavir, but the result (CYP3A4 inhibition with P-gp induction) does not seem likely to increase colchicine serum concentrations based on study of tipranavir/ritonavir with other CYP3A4/P-gp substrates [71] kBoth colchicine and cyclosporine can cause myopathy, and all case reports of colchicine–cyclosporine DDI had evidence of myopathy (e.g., some combination of muscle pain, muscle weakness, dark urine) lThe lack of effect of grapefruit juice in the healthy subject study is consistent with the theory that inhibition of both CYP3A4 and P-gp is necessary for a DDI with colchicine. It is possible that the huge amount of grapefruit juice ingested by the 8-year-old girl (1 L/day for 2 months) was enough to inhibit P-gp. It is also possible that the grapefruit juice used in the healthy subject study did not contain sufficient quantities of the CYP3A4-inhibiting phytochemicals to affect colchicine pharmacokinetics. Different types of grapefruit are known to contain differing amounts of CYP3A4-inhibiting compounds [94, 95] mAtorvastatin, lovastatin, and simvastatin are all metabolized by CYP3A4, so patients on CYP3A4 inhibitors are likely to be at increased risk nPitavastatin, pravastatin, and rosuvastatin are all substrates for OATP transporters, so patients on OATP inhibitors may be at increased risk Risk Factors for Colchicine DDIs The risk of adverse outcomes from colchicine DDIs can be influenced substantially by the dose and duration of colchicine therapy as well as kidney and liver function Colchicine Dose and Duration Epidemiological studies using low-dose colchicine have generally found little evidence of adverse consequences with colchicine and interacting medications [1–4]. This is consistent with the case reports described in the online Supplemental Table in which the majority of adverse DDIs reported involved colchicine doses >0.5 mg/day (see ESM). Nonetheless, the epidemiological studies were not designed to detect adverse DDIs and we cannot rule out that some at-risk patients had DDIs. At colchicine doses > 0.5 mg/day, as one would expect, the risk of colchicine adverse DDIs appears to increase as the colchicine dose is increased. The duration of colchicine use is also important. It is unusual to see significant colchicine toxicity until at least 3 or 4 days of concurrent use of colchicine and the interacting drug, although in rare cases it can occur after only a day or two of concurrent therapy [18]. The case reports in the online Supplemental Table also provide onset of adverse reaction information. A more typical onset of adverse effects from the colchicine DDI would be after 5 or 10 days of concurrent therapy, and in some cases (especially DDIs causing myopathy) it may take weeks or even months. Renal Disease/Age Renal impairment predisposes to colchicine toxicity [42]. In 21 patients with varying levels of kidney impairment, the three patients with low estimated glomerular filtration rate (eGFR) (mean eGFR = 25) had 51% higher colchicine AUC compared with six patients with normal GFR (mean eGFR = 110). In patients on hemodialysis three times a week with eGFR = 0, colchicine AUC was more than 5-fold higher than those with normal GFR [16]. Another study found a colchicine half-life to be four times higher in patients with renal insufficiency compared with those with normal kidney function [26]. A study of colchicine pharmacokinetics found total body clearance of colchicine about twice as high in younger subjects as in elderly patients with a mean creatinine clearance (CrCl) of 46 mL/min [22]. Elderly patients may also be at higher risk of colchicine toxicity due to polypharmacy. Renal impairment may also increase the risk of colchicine DDIs by increasing the serum concentration of the drug interacting with colchicine. For example, fluconazole displays dose-dependent inhibition of CYP3A4, and undergoes renal elimination, so patients with severe renal dysfunction would tend to have higher concentrations of fluconazole [43]. This would lead to increased CYP3A4 inhibition, as well as compromised renal elimination of colchicine, a combination that may have led to colchicine toxicity [44]. In many case reports of colchicine toxicity due to DDIs described in the online Supplemental Table (see ESM), renal impairment appeared to be a predisposing factor. It is not possible to precisely determine the degree of increased risk of colchicine DDIs at given levels of renal impairment, but a consensus statement on the use of colchicine used an eGFR of < 30 mL/min as the point where colchicine concentrations can start to increase substantially [10]. Based on the available data, we might estimate that the renal elimination of colchicine starts to be compromised somewhere around an eGFR of 60 mL/min and becomes a serious problem as one approaches an eGFR of 30 mL/min. Liver Disease As with many drugs, the effect of liver disease on colchicine elimination is not well characterized. There are many types of liver disease with different etiologies, and varying effects on drug metabolism and elimination. Moreover, there is no agreed upon measurement of liver function, unlike eGFR for kidney function, to quantify the likely reduction in colchicine elimination in patients with liver disease. One consensus statement on the use of colchicine in liver disease used a Child-Pugh score of C as the point where the half-life of colchicine starts to increase substantially [10], and, in the absence of definitive data, reduction in colchicine dose at this level of hepatic impairment appears to be a reasonable guideline. Nonetheless, patients with lesser degrees of hepatic malfunction may have some reduction in colchicine clearance, thus increasing the risk of colchicine DDIs to some degree. Transporter Polymorphisms Given that colchicine is a substrate for P-gp, it would be expected that those with polymorphisms resulting in low P-gp activity may be at increased risk for adverse colchicine DDIs. For example, drugs that primarily inhibit CYP3A4 and appear to have little effect on colchicine (e.g., voriconazole) might have a greater effect in patients with low P-gp activity due to genomics or other drugs. Little is currently known regarding the effect of transporters other than P-gp on colchicine pharmacokinetics, so it may be useful to study the effect of polymorphisms in MRP2 or OATPs on colchicine disposition. Dose/Duration of Perpetrator Drug The magnitude of perpetrator drug inhibition or induction on drug metabolizing enzymes and transporters can be affected by the dose of the drug and the duration of perpetrator drug therapy. Management Options Given the potentially life-threatening outcomes of colchicine drug interactions and the difficulty of treating severe colchicine toxicity once it occurs, it is imperative to avoid putting patients at risk. There are essentially three ways to reduce the risk of adverse outcomes from colchicine DDIs: (i) avoid using the interacting drug while the patient is on colchicine, (ii) avoid using colchicine during administration of the interacting drug, or (iii) give colchicine and the interacting drug, but reduce the colchicine dose. We now consider these three approaches. Avoid Using the Interacting Drug If the patient on colchicine has severe renal or hepatic impairment, it is particularly important to avoid the use of CYP3A4/P-gp inhibitors. But even when the patient does not have significant renal or hepatic disease, avoiding the interacting drug is likely to be the preferred option when it is feasible. For example, it is difficult to imagine a scenario in which it would be necessary to give clarithromycin to a patient on colchicine, a combination that has produced numerous fatalities [18]. If another antibiotic is suitable, it would almost always be preferable to use it in place of the clarithromycin (most antibiotics do not inhibit CYP3A4 and P-gp). A similar argument could be made for calcium channel blockers, where diltiazem and verapamil inhibit CYP3A4 and P-gp, but most of the other calcium-channel blockers do not. Other calcium-channel blockers may or may not be appropriate to use in place of diltiazem or verapamil in any given patient, but if they are, an interaction with colchicine could be easily avoided. Table 3 presents non-interacting alternatives for many of the drugs that can interact with colchicine. Stopping Colchicine During Use of the Interacting Drug If the interacting drug is clinically necessary, stopping colchicine during the use of the interacting drug may be appropriate, especially if the interacting drug is used for a limited time (such as an antibiotic). Whether this is a viable option would also depend on the indication for colchicine. If colchicine is being used for gout, short-term alternatives to colchicine may be available. If colchicine is used for familial Mediterranean fever or for cardiovascular disorders such as pericarditis or post-myocardial infarction, stopping colchicine may or may not be the best option. Reduce Colchicine Dose When Precipitant Drug is Started If concurrent use of colchicine with the interacting drug cannot be avoided—which should rarely be the case—one could consider prophylactic colchicine dosage reductions. Theoretically, reducing the colchicine dose when the precipitant drug is added would offset the pharmacokinetic interaction and maintain colchicine serum concentrations in the therapeutic range, as has been recommended in the labeling for some colchicine products [25] and in published articles [45, 46]. In practice, however, because the magnitude of these DDIs is so variable from one person to another, a ‘one size fits all’ a priori colchicine dosage reduction is likely to be excessive for some people (resulting in subtherapeutic colchicine concentrations) and insufficient for others (leading to colchicine toxicity). Accordingly, colchicine dosage reduction is best considered a last resort, and only when both colchicine and the interacting drug must be used together. Consider the interaction with ritonavir, where the mean increase in colchicine AUC following ritonavir was 296%, but the range was 54–924% as shown in Fig. 1 [25, 45]. Despite this marked variability, the colchicine labeling for Colcrys states that if colchicine and ritonavir are used together, the colchicine dose should be reduced by 50% [25]. (One possible source of ritonavir’s highly variable effect on colchicine is that ritonavir affects transporters in addition to P-gp, and affects CYP isozymes in addition to CYP3A4, but this hypothesis requires confirmation.)Fig. 1 The high variability of changes in colchicine AUC in 18–24 healthy subjects given a single dose of colchicine 0.6 mg with and without pretreatment with various drugs. The bars represent the subject with the smallest increase and the subject with the largest increase for each interacting drug. Adapted from data presented in references [25, 45, 47] The marked variability seen in healthy subjects is likely to be even greater in patients, who may be taking various other medications, may have varying degrees of renal or hepatic impairment, and may have other disorders that affect colchicine pharmacokinetics. Moreover, the much larger numbers in the patient population will virtually guarantee that some patients will have pharmacogenomic differences that will predispose them to colchicine toxicity (e.g., low P-gp activity). Finally, the pharmacokinetic studies conducted in healthy subjects (shown in Fig. 1) involved single, small doses of colchicine, unlike what is usually the case for patients receiving colchicine therapeutically. In patients who develop serious colchicine toxicity from a DDI, multiorgan failure can occur with marked reduction in kidney function; this in turn can worsen the colchicine toxicity. None of these problems would occur with single-dose studies in healthy subjects. For all of these reasons, one cannot use the single-dose studies in healthy subjects to predict whether a small or marked increase in colchicine plasma concentrations will occur in any given patient. Figure 2 represents a hypothetical estimate of what might be expected to happen if we follow the rule of thumb recommendations for avoiding colchicine toxicity when patients receive CYP3A4/P-gp inhibitors with colchicine. These are hypothetical scenarios but they are entirely plausible given the evidence of high intersubject variability in the magnitude of colchicine DDIs as shown in Fig. 1. In Fig. 2, ‘a’ represents the starting conditions in a group of patients, and to simplify the discussion let us assume that they all have therapeutic colchicine serum concentrations. Next is ‘b’, which represents what may happen if a CYP3A4/PGP inhibitor is added, but the DDI is ignored; here some patients are likely to develop colchicine toxicity and some may die. Next is ‘c’, in which we adopt the colchicine dosage reduction recommendations, but—due to the large variability—some patients become subtherapeutic, and others develop colchicine toxicity. Finally, ‘d’ represents what may happen if we lowered the colchicine dose sufficiently to avoid all colchicine toxicity, resulting in an unacceptable number of patients with subtherapeutic colchicine levels.Fig. 2 Hypothetical predicted outcomes for various ways of managing the addition of CYP3A4/P-gp inhibitors to colchicine. a Before interacting drug added, b interacting drug added with no change in colchicine dose, c colchicine dosage reduction method, d reducing colchicine dosage enough to avoid all colchicine toxicity Whether or not it is appropriate to prophylactically adjust the dose of an object drug when starting a precipitant drug depends on the drugs involved in the interaction. For some drugs, such as warfarin, if one adjusts the warfarin dose when adding a CYP2C9-inhibiting drug, at least one can monitor the INR to determine if the warfarin dosage adjustment was too large or too small. For colchicine, however, serum concentrations are not routinely available at this time, and life-threatening colchicine toxicity (e.g., pancytopenia, multi-organ failure, rhabdomyolysis) can begin after only a few days of administration of the interacting drug (see case reports in the online Supplemental Table, ESM). There is no specific remedy, and once the kidney and liver start to fail, colchicine may persist for weeks until the patient dies or (usually very slowly) recovers [48–50]. Dialysis is not effective in colchicine poisoning, but in one case the patient was successfully treated with kidney replacement therapy and plasmapheresis [51]. It is not known if giving an enzyme inducer such as rifampin would help eliminate colchicine, but it is an intriguing possibility given that rifampin probably increases colchicine elimination [31]. Rifampin has been used in poisonings with other drugs to increase elimination [52], but the efficacy and safety of using rifampin for colchicine toxicity is not established. The use of activated charcoal (and possibly other binding agents) may be worth trying, since colchicine undergoes enterohepatic circulation, although this has not been studied [11]. A summary of suggested management options for colchicine DDIs with CYP3A4/P-gp inhibtors is presented in Fig. 3.Fig. 3 Management option algorithm for minimizing the risk of colchicine drug–drug interactions Patient Education If concomitant use of colchicine with interacting drugs is necessary, it is crucial to advise the patient to be alert for evidence of colchicine toxicity. There are three primary types of toxicity for which the patient should be vigilant:Gastrointestinal toxicity: severe diarrhea (often with some combination of nausea, vomiting, or abdominal pain) Neuromuscular toxicity: muscle weakness, myalgia, dark or brown urine Hematologic toxicity (pancytopenia): fever, other signs of infection, excessive or unusual bleeding Mild to moderate diarrhea is common when colchicine serum concentrations are elevated, but if the diarrhea is severe or prolonged, the patient should contact their prescriber. Regarding signs and symptoms of myopathy, some have proposed a ‘classic triad’ of myalgia, muscle weakness, and ‘tea-colored urine’ [53]. However, muscle weakness often occurs without myalgia, and sometimes the reverse is true (see cases in the online Supplemental Table, ESM). Darkened urine from myoglobinuria is usually a late sign of myotoxicity, and often occurs after the muscle weakness and/or myalgia has become so severe that the patient has already sought medical care. Moreover, using the term ‘tea-colored urine’ is problematic given that tea can have many different colors from dark brown to very light, so it would be better to say ‘dark’ or ‘brown’ urine. Darkened urine is a particularly ominous sign, because the excessive myoglobinuria may lead to acute kidney injury, thus prolonging the colchicine toxicity. Colchicine toxicity can produce many other signs and symptoms, but they are either relatively nonspecific (fatigue, lethargy, malaise, insomnia), or they occur relatively infrequently (shortness of breath, cough, cardiac arrhythmias), or they tend to occur late in the course of colchicine toxicity (alopecia). Accordingly, it is probably best to concentrate the patient education on the more specific and common signs and symptoms as described above. Drugs That May Interact with Colchicine Colchicine DDIs have been extensively studied, and include pharmacokinetic studies, case reports, and case series. See Table 3 for a summary of these DDIs as well as management recommendations. Details of all case reports mentioned in Table 3, including supporting references, can be found in the online Supplemental Table (see ESM). Misleading Colchicine Drug Interaction Information Colchicine DDI information for health professionals is available from a variety of sources. The medical literature provides numerous pharmacokinetic studies, case reports, epidemiological studies, and reviews on colchicine DDIs, resulting in many conflicting recommendations. There are also numerous inconsistencies in the prescribing information for the various colchicine products, as well as in online DDI checking software and in clinical decision support systems [3, 10, 25, 30, 45, 46, 54–57]. The result is a jumble of confusion, with errors of omission, errors of commission, conflicting statements, and vague pronouncements as summarized in Table 4.Table 4 Misleading statements in published literature, product information, online sources [3, 10, 25, 30, 45, 46, 54–57] Type of problem Examples Recommending colchicine dosage adjustments instead of avoiding combinations For the reasons detailed in the previous sections, if the drug interaction with colchicine is important enough to warrant a dose reduction, it is usually safer to avoid the combination entirely if possible Errors of omission Virtually all sources have incomplete lists of drugs likely to interact with colchicine (e.g., CYP3A4/P-gp inhibitors, enzyme inducers) Errors of commission Many sources list drugs as interacting with colchicine that are unlikely to affect colchicine AUC or toxicity (for explanations, see Table 3). Errors include: Voriconazole as increasing colchicine AUC Tipranavir as increasing colchicine AUC Digoxin as a “potentially significant drug interaction,” a ‘serious’ DDI, and as contraindicated Errors regarding effects of drugs on enzymes and transportersa Voriconazole as a P-gp inhibitor Clarithromycin as a ‘moderate’ 3A4 inhibitor Fluconazole as a strong P-gp inhibitor Cobicistat as a ‘mild’ CYP3A4 inhibitor Erythromycin as a ‘mild’ CYP3A4 inhibitor Amiodarone as a ‘mild’ P-gp inhibitor; no mention of 3A4 Diltiazem and verapamil as ‘mild’ P-gp inhibitors, with no mention of 3A4 Ritonavir as a ‘moderate’ inhibitor of CYP3A4 and P-gp Quinidine as a ‘mild’ P-gp inhibitor Cyclosporine as a ‘moderate’ CYP3A4 inhibitor, but no mention of P-gp Excessive reliance on purported magnitude of inhibitory potency on CYP3A4 and P-gp (strong, moderate, mild) The desire to have set guidelines is understandable, but these categories of inhibitors provide general recommendations. Clinical consequences may vary based on a host of other factors, only some of which have been characterized in clinical studies. There is a large overlap in the magnitude of effect of ‘moderate’ and ‘strong’ inhibitors in PK studiesb Failure to consider dose of inhibitor Fluconazole and grapefruit both inhibit CYP3A4, but the magnitude of inhibition is dose-related. Small doses usually have little effect, while large doses may have a strong inhibitory effect on CYP3A4c Vague statements Some sources list HIV medications (ritonavir) under moderate P-gp inhibitors. This is misleading because: (1) Some HIV medications are enzyme inducers and would be expected to reduce colchicine concentrations (e.g., tipranavir, efavirenz, etravirine, nevirapine), and (2) Many HIV medications other than ritonavir also inhibit CYP3A4/P-gp, but they are not listed as interacting with colchicine Excessive reliance on data from epidemiological studies that were not designed to assess colchicine DDIs Epidemiological studies of colchicine for coronary artery disease  Used low doses of colchicine (usually 0.5 mg/day)  Patients were screened for risk factors (renal disease, drugs, etc.)  Often relied on self-reporting of ADRs by patient  Laboratory confirmation of DDIs may or may not have been done Epidemiologic studies for colchicine DDIs (e.g., Kwon statin study)  Unlikely to detect rare DDIs due to inadequate power  Statins were lumped together to obtain results ADRs adverse drug reactions, AUC area under the concentration–time curve, DDIs drug–drug interactions, P-gp P-glycoprotein, PK pharmacokinetics aErrors partly result from relying on in vitro or animal studies without human in vivo data bFor example, as seen in Fig. 1, the ‘moderate’ CYP3A4 inhibitor diltiazem produced a 339% increase in colchicine AUC in one healthy subject, while the ‘strong’ CYP3A4 inhibitor ritonavir produced a 54% increase in colchicine AUC in another subject cOther factors may also contribute to the variability in fluconazole effect on colchicine, such as low P-gp activity (due to genetics or drugs) or serious renal impairment (increasing concentrations of both fluconazole and colchicine) A striking example is the variability in recommendations for using concurrent colchicine and statins. Some have claimed that combining colchicine and statins does not cause adverse DDIs [46, 54] while others urge that the combinations be avoided [55]. A thorough assessment of the data by Wiggins et al. fell between these extremes and provided nuanced and useful guidelines [58], and a systematic review of the colchicine-statin DDI thoroughly assessed the data and concluded that the DDIs can be serious, and mitigation strategies are necessary to avoid patient harm [59]. While the data suggest that most people using concurrent statins and low-dose colchicine do not develop myopathy, reasonable precautions are warranted as discussed in Table 3. In addition to improving the information on colchicine DDIs, it may be necessary to raise awareness in health care professionals on how serious these DDIs can be. In one study, over 90% of the alerts for colchicine with strong CYP3A4 inhibitors flagged in clinical decision support systems were overridden by prescribers [60]. It seems likely that more than 10% of these patients were at risk, and these results suggest a lack of awareness regarding the potentially life-threatening nature of colchicine DDIs. Conclusion Colchicine is a useful drug for a wide range of disorders and is usually safe when used in appropriate doses given the patient’s renal function. Numerous drugs are capable of causing colchicine toxicity, however, and given that colchicine toxicity is potentially fatal and difficult to treat, it is imperative that every effort be made to avoid placing patients at risk from these drug interactions. Currently available information on colchicine DDIs can be confusing and inconsistent, and in this paper we have tried to present recommendations based on the empirical evidence. Enough is known about colchicine drug interactions so that virtually every case of colchicine toxicity could have been prevented had the appropriate precautions been taken. Supplementary Information Below is the link to the electronic supplementary material.Supplementary file1 (PDF 631 KB) Declarations Funding This project was supported by grant R01HS025984 from the Agency for Healthcare Research and Quality. The content is solely the responsibility of the authors and does not necessarily represent the official views of the Agency for Healthcare Research and Quality. Conflict of interest Philip Hansten and John Horn receive royalties from books published on drug–drug interactions. There was no support from any organization for the submitted work. None of the authors have a financial relationship with any organizations that might have an interest in the submitted work in the previous 3 years, and have no other relationships or activities that could appear to have influenced the submitted work. Consent to participate Not applicable. Consent for publication Not applicable. Consent to participate Not applicable. Availability of data and material Not applicable. Code availability Not applicable. Author contributions This paper arose out of the discussions of a research team working on an AHRQ grant. We are studying risk factors for drug interactions, and we discussed colchicine drug interactions frequently over many months at our bi-weekly meetings. PDH drafted the manuscript. All authors contributed to critical revision of the manuscript, and approved the final version of the manuscript. The corresponding author attests that all listed authors meet authorship criteria and that no others meeting the criteria have been omitted. Ethical approval Not required. ==== Refs References 1. Chen K Schenone AL Borges N Militello M Menon V Teaching an old dog new tricks: colchicine in cardiovascular medicine Am J Cardiovasc Drugs 2017 17 347 360 10.1007/s40256-017-0226-3 28353024 2. Tardif J-C Kouz S Waters DD Bertrand OF Diaz R Maggioni AP Efficacy and safety of low-dose colchicine after myocardial infarction New Engl J Med. 2019 381 26 2497 2505 10.1056/NEJMoa1912388 31733140 3. Imazio M Nidorf M Colchicine and the heart Eur Heart J 2021 42 28 2745 2760 10.1093/eurheartj/ehab221 33961006 4. Nidorf SM Layland J Robinson PC Patel S Psaltis PJ Thompson PL Emerging evidence for the use of colchicine for secondary prevention of coronary heart disease Med J Austral. 2022 216 8 385 387 10.5694/mja2.51488 35393699 5. Toro-Huamanchumo CJ Benites-Meza JK Mamini-García CS Bustamante-Paytan D Gracia-Ramos AE Diaz-Vélez C Efficacy of colchicine in the treatment of COVID-19 Patients: a systemic review and meta-analysis J Clin Med 2022 11 9 2615 10.3390/jcm11092615 35566737 6. Mehta KG Patel T Chavda PD Patel P Efficacy and safety of colchicine in COVID-19: a meta-analysis of randomized controlled trials RMD Open 2021 7 e001746 10.1136/rmdopen-2021-001746 34810227 7. Ledford H The long wait for a long-COVID therapy Nature 2022 608 258 260 10.1038/d41586-022-02140-w 35945375 8. Tobler DL Pruzansky AJ Naderi S Ambrosy AP Slade JJ Long-term cardiovascular effects of COVID-19: emerging data relevant to the cardiovascular clinician Curr Atheroscl Rep. 2022 24 563 570 10.1007/s11883-022-01032-8 9. Stewart S Adverse events during oral colchicine use: a systematic review and meta-analysis of randomised controlled trials Arthritis Res Ther 2020 22 1 28 10.1186/s13075-020-2120-7 32054504 10. Robinson PC Terkeltaub R Pillinger MH Shah B Karalis V Karatza E Consensus statement regarding the efficacy and safety of long-term low-dose colchicine in gout and cardiovascular disease Am J Med 2022 135 32 38 10.1016/j.amjmed.2021.07.025 34416165 11. Finkelstein Y Colchicine poisoning: the dark side of an ancient drug Clin Toxicol 2010 48 407 414 10.3109/15563650.2010.495348 12. Garrouste C Philipponnet C Kaysi S Enache I Tiple A Heng AE Severe colchicine intoxication in a renal transplant recipient on cyclosporine Transplant Proc 2012 44 2851 2852 10.1016/j.transproceed.2012.09.028 23146540 13. Tröger U Lins H Scherrmann J-M Wallesch C-W Bode-Böger SM Tetraparesis associated with colchicine is probably due to inhibition by verapamil of the P-glycoprotein efflux pump in the blood-brain barrier BMJ 2005 331 613 10.1136/bmj.38568.639688.F7 16148013 14. Caraco Y Putterman C Rahamimov R Ben-Chetrit E Acute colchicine intoxication—possible role of erythromycin administration J Rheumatol 1992 19 494 496 1578471 15. Çelebi ZK Akturk S Oktay EI Duman N Keven K Colchicine-induced rhabdomyolysis following concomitant use of clarithromycin in a hemodialysis patient with familial Mediterranean fever Clin Kidney J 2013 6 665 666 10.1093/ckj/sft129 26120465 16. Amanova A Celebi ZK Bakar F Caglayan MG Keven K Colchicine levels in chronic kidney diseases and kidney transplant recipients using tacrolimus Clin Transplant 2014 28 1177 1183 10.1111/ctr.12448 25125128 17. Dogukan A Oymak FS Taskapan H Güven M Tokgoz B Utas C Acute fatal colchicine intoxication in a patient on continuous ambulatory peritoneal dialysis (CAPD). Possible role of clarithromycin administration Clin Nephrol 2001 55 181 182 11269688 18. Villa Zapata L Hansten PD Horn JR Boyce RD Gephart S Subbian V Evidence of clinically meaningful drug–drug interaction with concomitant use of colchicine and clarithromycin Drug Saf 2020 43 661 668 10.1007/s40264-020-00930-7 32274687 19. Hung IFN Wu AKL Cheng VCC Tang BSF To KW Yeung CK Fatal interaction between clarithromycin and colchicine in patients with renal insufficiency: a retrospective study Clin Infect Dis 2005 41 291 300 10.1086/431592 16007523 20. Tateishi T Soucek P Caraco Y Guengerich FP Wood AJ Colchicine biotransformation by human liver microsomes. Identification of CYP3A4 as the major isoform responsible for colchicine demethylation Biochem Pharmacol 1997 53 111 116 10.1016/s0006-2952(96)00693-4 8960070 21. Niel E Scherrmann J-M Colchicine today Jt Bone Spine. 2006 73 672 678 10.1016/j.jbspin.2006.03.006 22. Rochdi M Sabouraud A Girre C Venet R Scherrmann JM Pharmacokinetics and absolute bioavailability of colchicine after i.v. and oral administration in healthy human volunteers and elderly subjects Eur J Clin Pharmacol 1994 46 351 354 10.1007/BF00194404 7957521 23. Dahan A Sabit H Amidon GL Multiple efflux pumps are involved in the transepithelial transport of colchicine: combined effect of P-glycoprotein and multidrug resistance-associated protein 2 leads to decreased intestinal absorption throughout the entire small intestine Drug Metab Disposit. 2009 37 2028 2036 10.1124/dmd.109.028282 24. Wason S DiGiacinto JL Davis MW Effects of grapefruit and Seville oranges on the pharmacokinetic properties of colchicine in healthy subjects Clin Ther 2012 34 2161 2173 10.1007/s11095-008-9789-7 22940371 25. Colcrys (colchicine) tablet [prescribing information]. Deerfield, IL: Takeda Pharmaceuticals America, 2020. 26. Ben Chetrit E Scherrmann J-M Zylber-Katz E Levy M Colchicine disposition in patients with familial Mediterranean fever with renal impairment J Rheumatol 1994 21 710 713 8035398 27. Brüggemann RJM, Alffenaar J-WC, Blijlevens NMA, Billaud EM, Kosterink JGW, Verweij PE, et al. Clinical relevance of the pharmacokinetic interactions of azole antifungal drugs with other coadministered agents. Clin Infect Dis. 2009;48:1441–58. 10.1086/598327 28. Fihlman M Hemmilä T Hagelberg NM Backman JT Laitila J Laine K Voriconazole greatly increases the exposure to oral buprenorphine Eur J Clin Pharmacol 2018 74 1615 1622 10.1007/s00228-018-2548-8 30167757 29. Purkins L Wood N Kleinermans D Nichols D Voriconazole does not affect the steady-state pharmacokinetics of digoxin Br J Clin Pharmacol 2003 56 Suppl 1 45 50 10.1046/j.1365-2125.2003.01998.x 14616413 30. CEDR: Center for Drug Evaluation and Research, Application No. 204820Orig1s000, November 11, 2013, Hikma Pharmaceuticals. 31. Koo HS Kim S Chin HJ A case of high dose colchicine with no efficacy in a patient with chronic kidney disease taking rifampicin J Rheum Dis 2014 21 6 2014 10.4078/jrd.2014.21.6.314 32. Kazmi H Cauchi M Sub-therapeutic colchicine levels in a patient treated for pericarditis in the setting of carbamazepine use: a case report J Clin Pharm Ther 2021 00 1 3 10.1111/jcpt.13549 33. Shen Z Tieu K Wilson D Bucci G Gillen M Lee C Evaluation of pharmacokinetic interactions between lesinurad, a new selective urate reabsorption inhibitor, and commonly used drugs for gout treatment Clin Pharmacol Drug Dev. 2017 6 4 377 387 10.1002/cpdd.323 28074640 34. Pirzada NA Medell M Ali II Colchicine induced neuromyopathy in a patient with normal renal function J Clin Rheumatol 2001 7 374 376 10.1097/00124743-200112000-00005 17039178 35. Chattopadhyay I Shetty HGM Routledge PA Jeffery J Colchicine induced rhabdomyolysis Postgrad Med J 2001 77 191 192 10.1136/pmj.77.905.191 11222829 36. Caglar K Odabasi Z Safali M Yenicesu M Vural A Colchicine-induced myopathy with myotonia in a patient with chronic renal failure Clin Neurol Neurosurg 2003 105 274 276 10.1016/s0303-8467(03)00030-1 12954545 37. Sayarlioglu M Sayarlioglu H Ozen S Erkoc R Gul A Colchicine-induced myopathy in a teenager with familial Mediterranean fever Ann Pharmacother 2003 37 1821 1824 10.1345/aph.1D188 14632592 38. Wilbur K Makowsky M Colchicine myotoxicity: case reports and literature review Pharmacotherapy 2004 24 1784 1792 10.1592/phco.24.17.1784.52334 15585444 39. Melli G Chaudhry V Cornblath DR Rhabdomyolysis: an evaluation of 475 hospitalized patients Medicine (Baltimore) 2005 84 377 385 10.1097/01.md.0000188565.48918.41 16267412 40. Altman A Szyper-Kravitz M Shoenfeld Y Colchicine-induced rhabdomyolysis Clin Rhematol. 2007 26 2197 2199 10.1007/s10067-007-0682-2 41. Gilibili RR Chatterjee S Bagul P Mosure KW Murali BV Mariappan TT Coproporphyrin-I: a fluorescent, endogenous optimal probe substrate for ABCC2 (MRP2) suitable for vesicle-based MRPT inhibition assay Drug Metab Dospos. 2017 45 604 611 10.1124/dmd.116.074740 42. Kuncl RW Duncan G Watson D Alderson K Rogawski MA Peper M Colchicine myopathy and neuropathy N Engl J Med 1987 316 1562 1568 10.1056/NEJM198706183162502 3035372 43. Berl T Wilner KD Gardner M Hansen RA Farmer B Baris BA Pharmacokinetics of fluconazole in renal failure J Am Soc Nephrol 1995 6 242 247 10.1681/ASN.V62242 7579091 44. Su Y-C Wu C-C Colchicine-induced acute neuromyopathy in a patient using concomitant fluconazole: case report and literature review Drug Saf Case Rep 2015 2 16 10.1007/s40800-015-0020-6 27747728 45. Terkeltaub RA Furst DE Digiacinto JL Kook KA Davis MW Novel evidence-based colchicine dose-reduction algorithm to predict and prevent colchicine toxicity in the presence of cytochrome P450 3A4/P-glycoprotein inhibitors Arthritis Rheum 2011 63 2226 2237 10.1002/art.30389 21480191 46. Vrachatis DA Papathanasiou KA Giotaki SG Iliodromitis KE Papaionnaou TG Repurposing colchicine’s journey in view of drug-to-drug interactions. A review Toxicol Rep 2021 8 1389 1393 10.1016/j.toxrep.2021.07.009 34285885 47. Wason S DiGiacinto JL Davis MW Effect of cyclosporine on the pharmacokinetics of colchicine in healthy subjects Postgrad Med 2012 124 4 189 196 10.3810/pgm.2012.07.2579 22913907 48. Eleftheriou G Bacis G Fiocchi R Sebastiano R Colchicine-induced toxicity in a heart transplant patient with chronic renal failure Clin Toxicol 2008 46 827 830 10.1080/15563650701779703 49. Bouquié R Deslandes G Renaud C Dailly E Haloun A Jolliet P Colchicine-induced rhabdomyolysis in a heart/lung transplant patient with concurrent use of cyclosporine, pravastatin, and azithromycin J Clin Rheumatol 2011 17 28 30 10.1097/RHU.0b013e3182056042 21169852 50. Mullins M Cannarozzi AA Bailey TC Ranganathan P Unrecognized fatalities related to colchicine in hospitalized patients Clin Toxicol 2011 49 648 652 10.3109/15563650.2011.589844 51. Schaffer DH Overbeek DL Erckson TB Boyer EW Goldfine C Muhsin SA Severe colchicine poisoning treated successfully with kidney replacement therapy and plasmapheresis: a case report Toxicol Commun. 2022 6 1 47 51 10.1080/24734306.2022.2055817 35497377 52. Unal S Bayrakci B Yasar U Karagoz T Successful treatment of propafenone, digoxin, and warfarin overdoseage with plasma exchange therapy and rifampicin Clin Drug Investig 2007 27 7 505 508 10.2165/00044011-200727070-00008 53. Torres PA Helmstetter JA Kaye AM Kaye AD Rhabdomyolysis: pathogenesis, diagnosis, and treatment Ochsner J 2015 15 58 69 25829882 54. Kwon OC Hong S Ghang B Kim YG Lee CK Yoo B Risk of colchicine-associated myopathy in gout: influence of concomitant use of statin Am J Med 2017 130 583 587 10.1016/j.amjmed.2016.12.006 28065770 55. Şen S Karahan E Büyükulaş C Polat YO Üresin AY Colchicine for cardiovascular therapy: A drug interaction perspective and a safety meta-analysis Anatol J Cardiol 2021 25 753 761 10.5152/AnatolJCardiol.2021.707 34734808 56. Mitigare (colchicine) capsule [prescribing information]. Memphis, TN: Hikma Americas, Inc.; June 2020. 57. Medscape Drug Interaction Checker, August, 2022. 58. Wiggins BS Saseen JJ Page RL II Reed BN Sneed K Kostis JB Recommendations for management of clinically significant drug-drug interactions with statins and select agents used in patients with cardiovascular disease. A scientific statement from the American Heart Association Circulation 2016 134 e468 e495 10.1161/CIR.0000000000000456 27754879 59. Schwier NC Cornelio CK Boylan PM A systematic review of the drug interaction between statins and colchicine: patient characteristics, etiologies, and clinical management strategies Pharmacotherapy 2022 42 320 333 10.1002/phar.2674 35175631 60. Van de Sijpe G Quintens C Walgraeve K van Laer E Penny J de Vlieger G Overall performance of a drug-drug interaction clinical decision support system: quantitative evaluation and end-user survery BMC Med Inform Decis Mak 2022 22 1 48 10.1186/s12911-022-01783-z 35193547 61. McDonald MG Au NT Rettie AE P450-based drug–drug interactions of amiodarone and its metabolites: diversity of inhibitory mechanisms Drug Metab Dispos 2015 43 1661 1669 10.1124/dmd.115.065623 26296708 62. Khan MH Rochlani Y Aronow WS Efficacy and safety of dronedarone in the treatment of patients with atrial fibrillation Expert Opin Drug Saf 2017 16 1407 1412 10.1080/14740338.2017.1387246 28960089 63. Wessler JD Grip LT Mendell J Giugliano RP The P-glycoprotein transport system and cardiovascular drugs J Am Coll Cardiol 2013 61 2495 2502 10.1016/j.jacc.2013.02.058 23563132 64. Bhardwaj R Collins JL Stingfellow J Madonia J Anderson MS Finley J P-glycoprotein and breast cancer resistance protein transporter inhibition by cyclosporine and quinidine on the pharmacokinetics of oral rimegepant in healthy subjects Clin Pharmacol Drug Dev. 2022 11 889 897 10.1002/cpdd.1088 35304977 65. Jerling M Huan B-L Leung K Chu N Abdallah H Hussein Z Studies to investigate the pharmacokinetic interactions between ranolazine and ketoconazole, diltiazem, or simvastatin during combined administration in healthy subjects J Clin Pharmacol 2005 45 422 433 10.1177/0091270004273992 15778423 66. Hylton AC Ezekiel TO Rhabdomyolysis in a patient receiving ranolazine and simvastatin Am J Health Syst Pharm 2010 67 1829 1831 10.2146/ajhp090299 20966146 67. Gumbrevičius G Damulevičienė G Galaunė V Gumbrevičiūtė M Possible interaction between dabigatran and ranolazine in patients with renal failure Medicina (Kaunas) 2019 56 13 10.3390/medicina56010013 31905748 68. Brings A Lehmann M-L Foerster KI Burhenne J Weiss J Haefeli WE Perpetrator effect of ciclosporin (P-glycoprotein inhibitor) and its combination with fluconazole (CYP3A4 inhibitor) on the pharmacokinetics of rivaroxaban in healthy volunteers Br J Clin Pharmacol 2019 85 1528 1537 10.1111/bcp.13934 30912163 69. Niwa T Imagawa Y Yamazaki H Drug interactions between nine antifungal agents and drugs metabolized by human cytochromes P450 Curr Drug Metab 2014 15 651 679 10.2174/1389200215666141125121511 25429674 70. Nivoix Y Levêque D Herbrecht R Koffel J-C Beretz L Ubeaud-Sequier G The enzymatic basis of drug- drug interactions with systemic triazole antifungals Clin Pharmacokinet 2008 47 779 792 10.2165/0003088-200847120-00003 19026034 71. Mukwaya G Interaction of ritonavir-boosted tipranavir with loperamide does not result in loperamide-associated neurologic side effects in healthy volunteers Antimicrob Agents Chemother 2005 49 4903 4910 10.1128/AAC.49.12.4903-4910.2005 16304151 72. Mora-Peris B Elso L Goldmeier D Mears A Weston R Cooke G A phase I study to assess the safety, tolerability and pharmacokinetic profile of boceprevir and sildenafil when dosed separately and together, in healthy male volunteers J Antimicrob Chemother 2015 70 1812 1815 10.1093/jac/dkv035 25693997 73. Marciano KA Seago K Dillaman M Ross KG Veltri L Cumpston A Evaluation of the pharmacokinetic interaction between letermovir and tacrolimus in allogeneic hematopoietic cell transplant recipients Transplant Cell Ther. 2022 S2666–6367 22 01194 10.1016/j.jtct.2022.03.028 74. Marshall WL McCrea JB Macha S Menzel K Liu F van Schanke A Pharmacokinetics and tolerability of letermovir coadministered with azole antifungals (posaconazole or voriconazole) in healthy subjects J Clin Pharmacol 2018 58 7 897 904 10.1002/jcph.1094 29578577 75. Beaudreuil S Schermann JM Verstuyft C Barrail-Tran A Smirnova M Durrbach A Differential pharmacokinetic interaction of cyclosporine and tacrolimus with colchicine in renal allograft recipients Clin Transplant 2018 32 e13405 10.1111/ctr.13405 30298530 76. Dahan A Amidon GL Grapefruit juice and its constituents augment colchicine intestinal absorption: potential hazardous interaction and the role of P-glycoprotein Pharm Res 2009 26 883 892 10.1007/s11095-008-9789-7 19048359 77. Davis MW Wason S Effect of steady-state atorvastatin on the pharmacokinetics of a single dose of colchicine in healthy adults under fasted conditions Clin Drug Investig 2014 34 259 267 10.1007/s40261-013-0168-8 78. Choi D-H Chung J-H Choi J-S Pharmacokinetic interaction between oral lovastatin and verapamil in healthy subjects: role of P-glycoprotein inhibition by lovastatin Eur J Clin Pharmacol 2010 66 285 290 10.1007/s00228-009-0757-x 20012601 79. Shah T McCarthy M Nasir I Archer H Ragheb E Kluger J Design and rationale of the colchicine/statin for the prevention of COVID-19 complications (C OLSTAT) trial Contemp Clin Trials 2021 110 106547 10.1016/j.cct.2021.106547 34461322 80. Holtzman CW Wiggins BS Role of P-glycoprotein in statin drug interactions Pharmacotherapy 2006 26 1601 1607 10.1592/phco.26.11.1601 17064205 81. Topliss D Gold M Kotsirilos V Lander C McNeil J Pillans P Fatal interactions and reactions with colchicine: beware CYP3A4 inhibitors Aust Adverse Drug React Bull 2008 27 5 18 82. Posada C García-Cruz A García-Doval I San Millán B Teijeira S Chloroquine-induced myopathy Lupus 2011 20 773 774 10.1177/0961203310385553 21300683 83. Aisen PS Marin DB Brickman AM Santoro J Fusco M Pilot tolerability studies of hydroxychloroquine and colchicine in Alzheimer disease Alzheimer Dis Assoc Disord 2001 15 96 101 10.1097/00002093-200104000-00009 11403336 84. Basaran O Aydin F Celikel BA Uncu N Cakar N Coexistence of systemic lupus erythematosus and familial Mediterranean fever in a pediatric patient Lupus 2016 25 1062 1063 10.1177/0961203315627872 26811369 85. Dabbagh MF Aurora L D’Souza P Weinmann AJ Bhargava P Basir MB Cardiac tamponade secondary to COVID-19 JACC Case Rep. 2020 2 1326 1330 10.1016/j.jaccas.2020.04.009 32328588 86. Rolle F Bincoletto V Gazzano E Coencapsulation of disulfiram and doxorubicin in liposomes strongly reverses multidrug resistance in breast cancer cells Int J Pharm 2020 580 119191 10.1016/j.ijpharm.2020.119191 32142738 87. Damkier P Hansen LL Brøsen K Effect of diclofenac, disulfiram, itraconazole, grapefruit juice and erythromycin on the pharmacokinetics of quinidine Br J Clin Pharmacol 1999 48 829 838 10.1046/j.1365-2125.1999.00099.x 10594487 88. Wu J Song Y Li H Chen J Rhabdomyolysis associated with fibrate therapy: review of 76 published cases and a new case report Eur J Clin Pharmacol 2009 65 1169 1174 10.1007/s00228-009-0723-7 19756555 89. Poruba M Matuskova Z Hüttl M Fenofibrate decreases hepatic P-glycoprotein in a rat model of hereditary hypertriglyceridemia Front Parmacol. 2019 7 10 56 10.3389/fphar.2019.00056 90. Gillen M Evaluation of pharmacokinetic interactions between lesinurad, a new selective urate reabsorption inhibitor, and CYP enzyme substrates sildenafil, amlodipine, tolbutamide, and repaglinide Clin Pharmacol Drug Dev. 2017 6 363 376 10.1002/cpdd.324 28067999 91. Harvey RD Morgan ET Cancer, inflammation, and therapy: effects on cytochrome P450-mediated drug metabolism and implications for novel immunotherapeutic agents Clin Pharmacol Ther 2014 96 449 457 10.1038/clpt.2014.143 24987833 92. Hansten PD Horn JR Hazlet TK ORCA: OpeRational ClassificAtion of drug interactions J Am Pharm Assoc 2001 41 2 161 165 10.1016/s1086-5802(16)31244-x 93. Foisy MM Yakiwchuk EM Hughes CA Induction effects of ritonavir: implications for drug interactions Ann Pharmacother 2008 42 1048 1059 10.1345/aph.1K615 18577765 94. Uesawa Y Mohri K Drug interaction potentials among different brands of grapefruit juice Pharmazie 2008 63 1440146 95. Guttman Y Yedida I Nudel A Zhmykhova Y Kerem Z Carmi N New grapefruit cultivars exhibit low cytochrome P4503A4-inhibition activity Food Chem Toxicol 2020 137 111135 10.1016/j.fct.2020.111135 31968226

36522578

PMC9754312

NO-CC CODE

2022-12-16 23:26:35

no

Drug Saf. 2022 Dec 15;:1-20

utf-8

Drug Saf

10.1007/s40264-022-01265-1

oa_other

==== Front Wireless Netw Wireless Networks 1022-0038 1572-8196 Springer US New York 3205 10.1007/s11276-022-03205-4 Original Paper Provably secure certificateless protocol for wireless body area network Mandal Susmita susmitamandal.nitrkl@gmail.com Dr. Susmita Mandal received her Ph.D. in computer science and engineering from National Institute of Technology Rourkela, India. She is currently an Assistant Professor associated with the Center for Distributed Ledger Technology and Innovation at Institute for Development and Research in Banking Technology, (Established by RBI), India. She is currently leading three Government sponsored Projects as Principle Investigator in the area of Cryptographic applications to Blockchain and secure communication using Internet of Things. She is the Managing Editor for the Journal of Banking and Financial Technology, Springer. Her current research interest are in Applied Cryptography, Security and Privacy aspects in Blockchain, Secure Low-cost communication solution, Authentication, and Privacy preserving mechanisms. grid.473631.4 0000 0004 1755 7841 Institute for Development and Research in Banking Technology, Hyderabad, India 15 12 2022 118 1 12 2022 © The Author(s), under exclusive licence to Springer Science+Business Media, LLC, part of Springer Nature 2022, Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. Wireless body area networks are gaining popularity due to their innovative applications such as timely analysis, remote monitoring of patients’ health, and high patient care quality. However, these healthcare systems that carry patient’s physiological data need special attention for the security and privacy of information. Due to the openness of transmitted data, the healthcare system gets prone to several adverse attacks. In this paper, a provably secure remote healthcare system is proposed based on the elliptic curve cryptosystem. The goal is to enable confidentiality and privacy of sensitive information by designing a certificateless authenticated key agreement protocol with low computational cost and higher security. The proposed scheme achieves anonymity, resistance to key escrow problems, mutual authentication between the sensor nodes attached to patients and the application provider. Furthermore, the protocol undergoes formal security analysis using the random oracle model, and the soundness of the proposed scheme is validated using ProVerif. Finally, the performance analysis depicts that the proposed scheme is efficient compared to existing methods. Keywords Authentication Certificateless Elliptic-curve cryptography ProVerif Wireless body area network (WBAN) ==== Body pmcIntroduction The rapid advancement in the Internet of Things (IoT) has brought significant improvements in human life. IoT enables a connection between interrelated computing devices with the Internet that gathers information over the network without any person-to-person or person-to-computer interaction. It has a broader application, like wireless sensor networks, smart homes, smart transportation, intelligent healthcare systems, etc. Among these, the wireless body area network (WBAN) has become an essential application in the healthcare ecosystem. WBANs are useful in short distance communication that consists of wearable sensor nodes responsible for monitoring the patient’s health-related sensitive information such as heartbeat rate, body temperature, blood pressure, blood sugar, oxygen level, etc. This technology provides a high quality of convenient and reliable service using IoT devices. These networks are beneficial to elderlies with permanent care at home. The biosensors are placed in or around the patient connected through a star or multi-hop topology. These sensors are responsible for sending the patient’s sensed data to the medical doctor to provide a real-time diagnosis with the right decisions. The shared information traverses several resource constraints devices, making it challenging to secure the transmitted data confidentiality. As if the patient’s physiological data is tampered with during the transmission process, it will mislead the physician, which will result in a false diagnosis of the patient’s health condition. Another crucial challenge concerns the resource-constrained devices connected to the patient; therefore, they must be exposed to lower complex computations for efficiency. Therefore, the patient’s medical record’s security and privacy are the primary concern in the healthcare industry. The data transmitted over the public network must be accessible by only authorized entities [1]. However, strong authentication and the key establishment must be achieved for securing the communication of WBAN. The first WBAN work was proposed by Zimmerman using a wireless personal area network (WPAN) technology [2]. In 2001, Van et al. introduced the concept of body area networks as a step towards a wearable future [3]. The traditional public-key cryptosystem uses trusted Certificate Authority (CA) to bind the user identity to the public key that causes heavy management overhead. Identity-based cryptosystem eliminates the need for explicit certificates by assigning public keys to its user identity; however, it suffers from the key-escrow problem. Over the decade, several WBAN models have been proposed. Still, the privacy and security of a patient stand as a big challenge for researchers. Authentication in WBAN is a relatively new research paradigm; however, few articles have recently discussed this research topic. Most of the existing schemes are based on traditional public key infrastructure (PKI) [4–6] and identity-based cryptosystem (IBC) [7–9]. Related work In the recent COVID-19 pandemic, the need for a remote health monitoring system shows promising solutions where a physician can remotely observe critical patients’ health status. However, patients’ physiological data need to be secured during transmission such that unauthorized entities can not access it. Therefore, it is necessary to enhance the security, which protects data from unauthorized manipulation and confidentiality to prevent data leakage. It is achieved from authenticated key agreement mechanism, which plays a vital role in dealing with the security requirements. Several schemes have recently been proposed for authenticating clients with the application provider remotely in a WBAN environment. However, these authentication schemes are based on traditional public-key cryptography (PKC) and identity-based cryptosystem (IBC) with complex computations. The difficulties in managing the certificates in public key infrastructure for the PKC make it unsuitable for WBAN. Whereas the IBC overcomes the certificate issuing and management problems, however, suffers from the key escrow problem. Al-Riyami and Paterson proposed a certificateless public key cryptography (CL-PKC) to overcome the issues mentioned earlier. However, the scheme increases the overall computation cost due to the usage of complex bilinear pairing operations [10]. In 2012, Drira et al. [11] has proposed an ID-based hybrid authentication and key establishment scheme based on a symmetric key cryptosystem. However, Kompara et al. [12] states that the protocol lacks data confidentiality, integrity, forward and backward secrecy. Also, it is susceptible to key escrow and impersonation attacks. In 2013, Liu et al. [13] proposed a lightweight certificateless authentication protocol based on a short certificateless signature method. However, the scheme fails to achieve session key security. Later Liu et al. [14] tried to resolve the issues mentioned in his above protocol by designing two certificateless remote anonymous authentication schemes for WBANs, namely, preliminary scheme and enhanced secure scheme. In contrast, Hu Xiong et al. [15] proves that the two protocols suffer from public key replacement attack. Zhao [16] pointed out that the preliminary version can not provide anonymity and the security-enhanced version suffers from stolen verifier-table attacks. In 2014, Zhao [16] proposed an efficient anonymous authentication scheme for wireless body area networks using ECC. Later, Wang et al. [17] demonstrated that [16] scheme lacks user anonymity and unable to provide unlikability features and proposed a new anonymous authentication scheme using bilinear pairing. In 2016, Wu et al. [18] found that [17] scheme is susceptible to impersonation attack. Recently, He et al. [19] proved that Liu et al. [13] also suffers from impersonation attack. Therefore, it may not suit the e-healthcare based privacy-preserving applications. Further, they have proposed a provably secure anonymous authentication scheme for WBANs. Several other schemes were proposed based on certificateless cryptosystem to overcome the traditional challenges, like Hu Xiong et al. [20] presented an anonymous certificateless authentication scheme for remote WBANs. Although the scheme withstands key escrow problems due to bilinear pairing usage, the scheme suffers from heavy computation overhead. Liu et al. [21] presented an anonymous 1-round authentication protocol for WBAN based on ECC and claims to achieve essential security features. However, Li et al. [22] prove that the scheme fails to provide key-compromise impersonation attack, stolen-verifier attack, and denial-of-service attack, and proposes an enhanced 1-round authentication protocol based on ECC. Later, Khan et al. [23] designed an improvement over Li et al. [22] by enabling a privacy-preserving key agreement for WBANs to achieve forward secrecy and unlinkability issues. Recently, Hassan et al. [24] proposed an ID-based authenticated key agreement protocol using a pairing-based cryptosystem. The protocol applies a ring signature to authenticate users within the multi-server environment anonymously. However, Kumar et al. [25] show that the scheme suffers from impersonation attack, man-in-the-middle attack, and significantly has higher computation cost. Shen et al. [26] proposed an anonymous certificateless authentication scheme. The protocol enables secure communication between hand-held PDA and application provider. However, the protocol lacks user anonymity and also suffers from collusion attack [27]. In 2020, Kasyoka et al. [28] proposed a pairing-free authentication scheme for healthcare management and proves that the protocol can thwart stolen verifier attacks. However, the scheme lacks rigorous formal security analysis. Recently, Sowjanya et al. [29] proved that [22] scheme lacks perfect forward secrecy, which is essential session key secrecy and has proposed a new end-to-end authenticated scheme for wearable monitoring devices. In the same year, Shuai et al. [30] introduced a privacy-preserving authentication scheme for WBANs using ECC suitable for multi-server architecture. Lately, Kumar et al. [31] proposed an identity-based anonymous authentication and key agreement scheme for WBAN. In 2021, Azees  et al. [32] proposed an efficient anonymous affine cipher-based encryption technique for WBANs. The work focuses on enhancing data confidentiality and authenticity, however the proposed model uses complex bilinear operations which increases the computational overhead. Therefore, the scheme may not be adequate for resource-constrained environment. Later, Lara  et al. [33] proposed a two-party authentication scheme using self-certified public keys based on ECC for healthcare application. The scheme focus to establish communication between the patient’s portable personal terminal and an application provider (AP) with a Two-party scheme. The scheme has lowered the computational cost but lacks consideration of honest but curious network manager during registration as the secret values are accepted without verification by end nodes. To address the high computation cost Soni  et al. [34] proposes an authentication and key agreement mechanism using low-cost functions (one-way hash, bit-wise XOR, and concatenation) for data exchanges in WBAN. The patient wearing a smart wearable device will collect real-time health information and share with healthcare providers. The protocol lacks discussion on prevention from hash collusion attack and is also vulnerable to offline password-guessing attacks. Peng  et al. [35] proposes an efficient certificateless online/offline signature scheme which is designed in a lightweight manner for WBANs. The scheme focuses on ensuring both security and efficiency of the online/offline signature for the real-world deployment. The scheme tries to reduce the computational cost by addressing the offline mode of verification. In order to achieve data confidentiality and fine-grained access control simultaneously on transmitted data Liu  et al. [36] proposed an attribute-based online/offline encryption and Identity-based ring signature scheme to achieve an outsourced online/offline hybrid signcryption mechanism applied for WBAN. The scheme allow patients to share fine-grained data without leaking any extra information. Despite its promising solution, the scheme may lead to the heavy computational cost on resource-constrained devices. Later, Cheng  et al. [37] proposed an improvement on Kumar et al. [31] scheme on lightweight cloud-assisted identity-based AKA scheme for WBAN. They claimed that the scheme lacks perfect forward secrecy and proposed a protocol a new anonymous identity-based AKA scheme. The proposed scheme claimed to be a certificateless AKA scheme, however the key shared by the network manager to cloud server and leaf node are not partial private keys but private keys. An approach of annonymization using identity-based authenticated encryption scheme without bilinear pairing, known as IB-AAE is proposed by Li  et al. [38]. The scheme combines the functionality of being anonymous and identity-based encryption, to achieve forward security. However, the generation of private keys are completely dependent on the trusted key generator. Recently, Hasan  et al. [39] proposed an architectural framework that incorporates blockchain with Software-Defined Wireless Body Area Networks (SDWBANs) to facilitate secure data sharing. The proposed framework of WBAN is modified by adding SDN enabled switches to communicate with sensors and forwarding the information through an interface between to Blockchain for just access validation. This solution may increase the overhead of data management and communication across the WBAN layers. So far, from the literature study, it is clear that using identity based cryptosystem may create a key escrow problem. As a malicious PKG could perform a man-in-the-middle (MITM) attack using the private keys. Therefore, desiging a certificateless authenticated key agreement protocol with backward and forward secrecy is suitable for resource constraint wireless body area networks. As the patient’s health information is very sensitive data, it must be accessed only by the authorized medical staff, including doctors and technicians. Therefore it is crucial to obtain data security to wireless body area networks such that confidential information may not be altered or abused by misusers. A remote WBAN based authenticated key agreement protocol must withhold the following properties: user authentication, data integrity, session key security, replay attack, impersonation attack, backward and forward secrecy. This paper proposes a provably secure certificateless authenticated key agreement protocol to meet the security requirements mentioned above and the challenges. The main contributions of this paper are summarized as follows: Design of a pairing-free secure authentication protocol that overcomes the traditional certificate issuing and management problem of public-key cryptosystem and achieves immunity against key escrow problem faced by IBC. The proposed scheme enables the network manager to generate partial private keys for each registered entity which can be validated publicly, thus preventing impersonating legitimate users. The security is based on the hardness assumption of Elliptic Curve Diffie–Hellman assumption and Computational Diffie–Hellman (CDH) problem. The scheme undergoes rigorous formal analysis and informal analysis using automated protocol analyzer ProVerif, and formal analysis using Real-Or-Random (ROR) model. The comparative analysis of the scheme is performed concerning computation, communication, and security features with existing schemes. Paper organization The rest of the paper is organized as follows. Section 2 deals with the mathematical background, network model, system model, and security model. The proposed scheme is depicted in Sect. 3. In Sect. 4, the formal informal security analysis along with protocol validation using ProVerif is presented. The performance analysis concerning security features, computation, and communication costs is shown in Sect. 5. Finally, we conclude in Sect. 6. Preliminaries This section provides brief introduction of cryptographic techniques used in this paper, network model, system model, and security model. Elliptic curve cryptography The security of Elliptic curve cryptography (ECC) is based upon the difficulty of solving Ellipic curve discrete logarithmic problem (ECDLP). Let E/Fq be a set of elliptic curve points over a finite field Fq, defined by an equation1 y2=x3+ax+b,a,b∈Fq where (4a3+27b2)≠0. The additive elliptic curve group defined as Gq={(x,y):x,y∈Fq,(x,y)∈E/Fq} ∪{O}, where the point “O” is known as “point at infinity” or “zero point”. The definitions about the elliptic curve group as follows.Point addition Let P, Q be two points on the curve shown in Eq. (2), such that P+Q=R, where the line joining P and Q intersects the curve at negative R, and the reflection towards x-axis is R. Scalar point multiplication It is defined on a cyclic group Gq as rP=P+P+⋯ +P(rtimes), where r∈Zq∗ is scalar. Computational problem Definition 1 (Elliptic curve discrete logarithm problem (ECDLP)) Given P,R∈Gq, where R=xP and x∈Zq∗. It is difficult to compute x from R. Definition 2 (Computational Diffie–Hellman problem (ECDH)) Given (P,xP,yP) ∈ Gq for x,y∈Zq∗, where computation of xyP is hard from the group Gq. Network model The WBAN ecosystem consists of in-body, on-body, and off-body sensors which communicate and share data across three layers. The description is depicted in Fig. 1.Layer 1 In this layer, the sensor nodes placed over and within the body communicate with the aggregator (i.e., mobile device). This layer is also known as Intra-BAN, i.e., an internal network. Layer 2 In this layer, the aggregator passes the collected data to the access points. This layer is also known as Inter-BAN. Layer 3 This layer depicts whole network of the server where communication happens beyond the BAN, therefore known as Beyond-BAN. The transmission occurs over a TCP/IP connection between the access points and the medical server. Fig. 1 Architecture of a WBAN System model The proposed model consists of three entities, namely, the Patient’s Mobile device (MD), Application Provider (AP), and Network Manager (NM). The model is depicted in Fig. 2.Patient’s mobile device (MD) The patient implies to the person who avails the medical facilities remotely. With sensors placed in or on the body to collect physiological information. These pieces of information are sent to an intermediate node known as an aggregator, such as PDA and hand-held mobile device. The sensors and mobile device should be registered with the Network Manager before it accesses the Application Provider’s services. Network manager (NM) It acts as a Key Generation Center (KGC) responsible for registering the sensor nodes, aggregator device, and application providers to legally access and avail the services. After the registration process, the NM generates partial private keys for every node and distributes them through a secure channel (i.e, TLS protocol). It is more likely a trusted third party that manages the whole network and participants. Application provider (AP) This represents the hospitals that the network manager authorizes to provide services to patients suffering from any critical ailment. Fig. 2 System model of a WBAN Security model This section outlines the widely accepted Dolev-Yao threat model  [40] pursued in the paper using the following assumptions:The Network Manager (NM) is assumed to be a trusted server that generates a partial private key for every registered user. Therefore, even if a passive/active adversary compromises the partial private key, he/she will not be able to forge the session key. The full private key is generated using the secret value and partial private key of each participating entity. The messages exchanged at the authentication phase between two entities are communicated over an insecure channel. An adversary can eavesdrop on all the messages transmitted and intercept, inject, modify, and resend any previously sent message. However, the adversary can not access messages in a secure channel. The application provider is assumed to be trustworthy; however, an adversary can compromise the database for malicious purposes. A privileged insider can act as an adversary by intercepting the registered request parameters. Proposed work In this section, the proposed certificateless authenticated key agreement protocol is discussed, which involves three phases: (1) Initialization, (2) Registration, and (3) Authentication. The registration occurs in a secure channel, with all the participating entities registering themselves with the network manager. A secure channel can be defined as a bidirectional communication medium that ensures the confidentiality, integrity, and freshness of data transferred through the channel. This can be achieved either by exchanging data through a trusted person in offline mode or through a strong Transport Layer Security (TLS) connection, defined in RFC 8446 [34, 41]. Typically the registration process is a one-time matter. Thus, an adversary cannot tamper the partial private keys sent by the entities during the registration process. In contrast, the authentication and key agreement phase between the aggregator and the application provider occur through an open/insecure channel, which means that an adversary (based on the Dolev-Yao model) can intercept, modify, delete the message tuple [42]. The notations used throughout the paper are mentioned in Table 1.Table 1 Notations Notation Description snm Master secret key of network manager Pnm Public key of network manager IDu Identity of patient xu Private key of patient Pu Public key of patient xap Private key of application provider Pap Public key of application provider ti/tu Timestamp P Generator of the elliptic curve || Concatenation function ⊕ XOR operation SK Session key between patient and application provider Hi Cryptographic hash function ∀i∈{0,1,2,3,4,5} A=?B Verifies whether A is equal to B Initialization phase Network manager chooses a security parameter 1k as input and generates a group G with prime order q and determines a point P as generator in group G. The NM then selects a random integer snm∈Zq∗ as a master key and computes a public key Pnm=(snm·P). Then six different hash functions are computed based on SHA-256 algorithm taking following three types of input sets: (a) {0,1}∗ is the set of binary bit-strings of arbitrary size, (b) Zq∗ is a set of positive integers where q is prime number, and (c) G is a cyclic multiplicative group of prime q, to obtain different hash values. The hash functions are depicted as follows: H0:{0,1}∗×Zq∗→Zq∗, where hash function H0 takes a set of binary bit-strings of arbitrary length concatenated with a set of integers and result is the integer coded in set Zq∗. Similarly, H1:Zq∗×Zq∗→Zq∗, H2:Zq∗×G→Zq∗, H3:{0,1}∗×G×G→Zq∗, H4:G×Zq∗×G×Zq∗→Zq∗, H5:{0,1}∗×{0,1}∗×{0,1}∗×G×G→Zq∗. Later, NM publishes the public parameters params={G,q,P,Pnm,H0,H1,H2,H3,H4,H5} while keeping the master key (snm) secret. Registration phase In this phase, the patients must register the mobile device with network manager. Similarly, the application provider must register with required details to the network manager. Where NM is a trusted authority. In real-time, the role of NM can be hosted by a central e-healthcare institution or any distributed authorized center. Patients willing to avail remote healthcare facilities from the application provider (i.e., medical institutions) must register their mobile devices assigned as an aggregator to receives details from the sensor nodes. The NM generates partial private keys to all the noted entities. The registration process is described below and also depicted in Fig. 3. Sensor nodes registration Step 1 The Application provider (AP) deploys the sensor nodes (SN) to each patient upon registration with NM. The AP generates IDSNi∈{0,1}∗ and a random secret sri∈Zq∗ where ∀i∈{1,2,3,…,n}. Then computes Ni=H0(sri||IDSNi) and sends the message tuple ⟨IDap,Ni⟩ to NM. Once received, NM stores it in its database. Similarly, AP sends ⟨IDSNi,Ni⟩ to sensor nodes later stores in its memory. Step 2 Upon receiving the message tuple NM generates a random number rSNi∈Zq∗, a fresh nonce Nc and computes Yi=Ni⊕H1(snm||rSNi), SKsn=H1(snm||rSNi)⊕(IDnm||Nc). Later after the registration of MD, network manager computes Ki=Ni⊕H2(Qmd||Rmd) and sends a message tuple ⟨Yi,SKsn,Nc,Ki,IDnm⟩ to sensor nodes. Step 3 Once SN receives the message, each sensor node computes H1(snm||rSNi)′=Ni⊕Yi. Then every node checks whether SKsn=?H1(snm||rSNi)′⊕(IDnm||Nc). If it matches then the sensor nodes successfully validates that NM has shared the correct parameter Ni for future communication else reply with an ⊥ message. Later, SN stores (Ki) in its database. Patients registration Step 1 Initially, the patient desiring to register for home care facility provides his/her identity, address proof along with device identification as IDmd∈{0,1}∗. Then generates a random number xmd∈Zq∗ and computes its respective public key Pmd=xmd·P. Then the patient’s device sends a message tuple ⟨IDmd,Pmd⟩ to NM. Step 2 Once received, NM first chooses a random number rmd∈Zq∗ then computes Rmd=rmd·P, Zmd=H3(IDmd||Pmd||Rmd), Qmd=(rmd+Zmd·snm) as partial private key. Then responds back with message ⟨IDnm,Ni,Qmd,Rmd⟩ to patient’s device. Step 3 Upon receiving the response, patient’s device computes Zmd′=H3(IDmd||Pmd||Rmd) and verifies the partial private key as Qmd·P=?(Rmd+Zmd′·Pnm). Therefore, patient’s mobile device holds the private keys Upriv=(xmd,Qmd) and public keys Upub=(Pmd,Qmd·P). Application provider registration Step 1 Like MD, the application provider sends his/her identity IDap∈{0,1}∗ then generates a random number xap∈Zq∗ and computes the public key Pap=xap·P. Then sends the message tuple ⟨IDap,Pap⟩ to NM. Step 2 Once received, NM first chooses a random number ra∈Zq∗ and a public key Ra=ra·P. Then computes Za=H3(IDap||Pap||Ra), Qa=(ra+Za·snm) as the partial private key. Then responds back with message ⟨IDnm,Qa,Ra⟩ to AP. Step 3 Upon receiving the message, the application provider then verifies the partial private keys as Za′=H3(IDap||Pap||Ra) then check if Qa·P=?Ra+Za′·Pnm. Therefore, AP holds private keys Apri=(xap,Qa) and public keys Apub=(Pap,Qa·P) respectively. Fig. 3 Registration phase Authentication phase Upon completing the registration process with the network manager, now each sensor node attached to the patient body, senses the health vitals and share it with the mobile device. The patient’s device is capable of communicating all the gathered information to the application provider for suitable diagnosis and treatment from concerned doctors. The process is described in following steps and depicted in Fig. 4.Step 1 To begin the communication, each sensor node collect the information related to blood glucose, temperature, oxygen levels, pulse, blood pressure etc. Then send the aggregated data along with a timestamp ti, parameter Ki stored in SN to the mobile device. The patient’s device first checks if |ti-tc|≤▵T to validate whether the received timestamp ti falls within the tolerable time delay ▵T else abort the message. Now MD computes Ni′=Ki⊕H2(Qmd||Pmd) then check if Ni′=?Ni from database. If matches, then aggregates the health vitals from all the nodes and generate a fresh nonce nu, timestamp tu, and an ephemeral key ymd∈Zq∗. Then computes, C1=H1(ymd||nu), F=xmd·Pap, C2=(Pnm||F||IDap||IDmd)⊕C1, C3=xmd+H4(Pap||tu||Pnm||Ni′)·C1. Later sends ⟨tu,C2,C3⟩ to the application provider. Step 2 Once received, AP first checks whether the time stamp tu is fresh as tu-tn≤▵T. Then computes F′=Pmd·xap, C1′=C2⊕(Pnm||F′||IDap||IDmd). AP checks whether C3·P=?Pmd+H4(Pap||tu||Pnm||Ni)·C1′·P if matches, then AP further generates a random number na and a time-stamp ta. Then computes, D1=(yap+na), Kz=C1′·Pap, D2=D1⊕(Kz||IDmd||IDap||C2), D3=xap+H4(C3||Pap||Pmd||ta)·D1. Finally, computes the session key SK=H5(C1′·D1·P||Kz||IDmd||IDap||IDnm). Later, sends a message tuple ⟨ta,D2,D3⟩. Step 3 Upon receiving the message tuple the patient’s device checks the freshness of the timestamp as ta-tn≤▵T. Then computes, Kz′=C1·Pap, D1′=D2⊕(Kz′||IDmd||IDap||C2). Patient’s device now checks whether D3·P=?Pap+H4(C3||Pap||Pmd||ta)·D1′·P if matches then patient computes the session key SK=H5(C1·D1′·P||Kz′||IDmd||IDap||IDnm) for future communication. Fig. 4 Authentication phase Security analysis In this section, a formal and informal (non-mathematical) security analysis of proposed scheme is performed. Furthermore, the protocol is verified using the widespread automated tool ProVerif. Formal proof using ROR model In this section, the formal security analysis using the probabilistic Real- Or-Random (ROR) model [43] is used to prove the session key security of the proposed scheme. The model states that an adversary A has complete control over all the transmitted messages between the entities such that, A can intercept, replay, or modify the messages. Though A does not have direct access to the private keys and session keys, however, it can perform the following queries to capture the leaked information. In this scheme, there are three participants Pdevice, Network Manager, and Application Provider. For convenience we denote Pdevice as Pi and Application Provider APj such that Pi and APj represents the ith and jth instances of Pdevice and AP in the authentication phase. Each instance is considered to be an oracle which has three states Accept, Reject, and ⊥, where Accept means oracle receives correct message else sends a Reject message otherwise send a ⊥ symbol means not able to produce a response. A can simulate following oracle queries:Execute(Pi,APj) It simulates passive attack and allow A to learn the messages exchanges between honest instances Pi and APj. Send(Pi/APj,m) It simulates active attack where A can generate any message m and send it to Pi/APj. As a result, the corresponding operation is performed according to the protocol description. SSReveal(Pi/APj) It allows A to obtain session-specific information. SKReveal(Pi/APj) It allows A to obtain the session key held by Pi/APj, if it has been negotiated. Corrupt(Pi/APj) This query is used to capture the perfect forward secrecy, in which A is allowed to obtain the long-term private key. Test(Pi/APj) This query returns a session key or a random value else sends back a null value. A is allowed to send a single Test query. In response a coin b∈{0,1} is flipped. If b=1 the session key is returned or a random value with same bit length is returned if b=0. Partnering The instances Pi and APj are partners if they authenticate each other and share the same session key. Freshness As instances, Pi and APj are fresh, if the session key is not revealed SKReveal. The adversary A’s goal is to identify the difference between a fresh session key from a random value. Semantic security An adversary A can execute several Test queries to either Pi or APj. In this query, the oracle flips the coin b, if a bit b′ is returned at the end of the experiment. A can win the game if b′=b. The advantage of A breaking the semantic security of the proposed certificateless authenticated key aggrement (CL-AKA) referred as W becomes AdvWCL-AKA(A)= 2Pr[b′=b]-1 where b′ is the bit A guesses. One-way hash function This query simulates the hash function. When Pi/APj receives the message m from A, the hash value of m is calculated and returned to A. Lemma 1 (Difference Lemma): Let R1,R2 and R3 denote the events defined in some probability distribution. If R1∧¬R3⇔R2∧¬R3, we have |Pr[R1]-Pr[R2]|≤Pr[R3] [44]. Theorem 1 Assume A is a probabilistic polynomial time adversary against the semantic security who can issue at most qs times Send query, qe time Execute query, and qh times hash query. The advantage of A is given as AdvWCL-AKA(A)≤(qh2/2(l+1))+(qs+qe)2/2p+(qs/2l)+qhAdvCL-AKAECDHP(A). Proof In order to prove the semantic security of the proposed scheme, a sequence of gamer Gm0 to Gm4 is presented where Gm0 represents the real attack. Let Succi is the event where the adversary (A) correctly guesses the bit b after the Test query. GameGm0 This games simulation is the real attack situation against the protocol in the random oracle model. Thus, we have 2 AdvWCL-AKA(A)=|2Pr[Succ0]-1| GameGm1 In this game, all the oracle queries and responses are stored in following list: LW stores all messages in the whole process. LH stores answers of all random hash oracles H0,H1,H2,H3,H4,H5. LT stores the transcript of all the messages. The answer to hash queries are generated in the form of (x,y,f) such that on a hash query f(x) where f∈{H0,H1,H2,H3,H4,H5}, if the record x,y,f is found in the list LH then return y directly, else a random string y with the same bit length will be produced as the returned value and then add x,y,f into the list LH. It is observed that the transcript distribution of games Gm0 and Gm1 are indistinguishable. Therefore, 3 Pr[Succ0]=Pr[Succ1] GameGm2 In this game, all the oracle queries are simulated as same as in the game Gm1, however collision occurred at transcript and collision occur at hash queries are aimed to be avoided. According to the birthday paradox, xmd,xap∈Zq∗, the probability of collision in the transcripts is at most (qs+qe)22p. The probability of hash collision is at most qh22l+1 where l is the length of hash output string. Therefore, we have |Pr[Succ2]-Pr[Succ1]|≤qh22l+1+(qs+qe)22p. GameGm3 In this game, if adversary (A) can guess the C3 and D3 without asking the random oracle queries H4, then the scheme is aborted. Such situation appears in the send queries, which means Gm3 and Gm2 are indistinguishable unless this case occurs. Thus, 4 |Pr[Succ3]-Pr[Succ2]|≤qs2l GameGm4 In this game, the session key security is considered. The notion of this security feature is that A must not be able to obtain the past session keys even if some information among {ymd,nu,yap,na,xap} is revealed. The adversary A knows the session transcripts (tu,C2,C3) and (ta,D2,D3). The adversary must ask H5 query to win the game. The goal of A is to compute the session key in the following four cases and by asking Execute(Pi,APj) and hash queries. (Case 1) Corrupt(Pi) and Corrupt(APj) are queried from which adversary A obtain the long-term private keys xmd,xap of Pi and APj respectively. However, to derive the session key SK=H1((ymd+nu)·(yap+na)·P||((ymd+nu)·xap·P)||IDmd||IDap||IDnm) either of the random nonces nu,na and the ephemeral key ymd of Pi and yap of APj are also required. (Case 2) SSReveal(Pi) and Corrupt(APj) are queried from which adversary A obtains the nonce nu, ephemeral keys ymd of Pi and long-term private key xap of APj. (Case 3) Corrupt(Pi and SSReveal(APj) are queried from which adversary A obtains the long-term secret key xmd of Pi and ephemeral key yap and nonce na. (Case 4) SSReveal(Pi) and SSReveal(APj) are queried from which adversary A obtains the ephemeral key of both but not the private key. However, in all the above four cases, the information available to adversary are insufficient to break the ECDHP assumption. As a result the difference between Gm3 and Gm4 is negligible as long as the ECDHP assumption holds. 5 |Pr[Succ4]-Pr[Succ3]|≤qhAdvCL-AKAECDHPA In Gm4, all the random oracles are simulated. A is only left to guess the winning bit b after querying the Test query. Therefore, we have,6 Pr[Succ4]=12 From Eq. 2, we have7 12AdvWCL-AKAA=|Pr[Succ0]-12| From Eqs. 3 and 4, we have8 12AdvWCL-AKAA=|Pr[Succ1]-12| Applying triangular inequality, we obtain,|Pr[Succ4]-Pr[Succ1]|≤|Pr[Succ4]-Pr[Succ3]|+|Pr[Succ3]-Pr[Succ1]|≤|Pr[Succ4]-Pr[Succ3]|+|Pr[Succ3]-Pr[Succ2]|+|Pr[Succ2]-Pr[Succ1]|≤qh22l+1+(qs+qe)22p+qs2l+qhAdvCL-AKAECDHP(A) From the games Gm0 to Gm4 and using the Lemma 1, Theorem 1 is proven. Informal security analysis Mutual authentication In the proposed scheme, at the authenticated phase the application provider verifies the authenticity of the patient communicating with his registered device as C3·P=?Pmd+H4(Pap||tu||Pnm||Ni)·C1′·P. Similarly, the patients device also verifies whether the response is obtained from legitimate application provider by checking if D3·P=?Pap+H4(C3||Pap||Pmd||ta)·D1′·P. Else the session is terminated. Therefore, the proposed scheme could provide mutual authentication successfully. Resistance against sensor node impersonation attack Suppose an adversary A intercepts the sensor node’s message ⟨ti,Ki⟩ claiming to be a legitimate sensor node to access the network for malicious gain. In the proposed scheme, an adversary fails to deduce the parameter Ki as the temporary identity Ni is computed using a random secret generated by the application provider and the identities of each sensor node. Then the one-way hash function is applied on Ni. Therefore, even if an adversary tries to compromise a sensor node the random secret cannot be disclosed. Perfect forward secrecy Suppose an adversary A had compromised the session key SK. The PFS holds when even after compromising long-term keys of the current session, it must not affect any past or future sessions. For instance, despite an adversary obtains Pdevice and AP’s private keys to compute the session key SK, an adversary A still requires the ephemeral keys and random nonces which are different and fresh at every session. Therefore, the proposed scheme achieves perfect forward secrecy. Resistance against application provider impersonation attack Suppose an adversary A intercepts the application provider message ⟨ta,D2,D3⟩ to forge its identity to the Patient’s device. However, in the proposed scheme, an adversary fails to deduce the parameter D1 as it includes a random ephemeral key yap and nonce na. Thus, it is difficult for A to obtain two secret parameters to forge successfully. Replay attack In the proposed scheme, the timestamp tu,ta is used between Pdevice and AP to prevent the replay attack. Even if an adversary tries to intervene the tolerable time delay will exceed and the session will be aborted. Therefore, it is infeasible to replay a message from any previous session into a new session. Resistance against patient impersonation attack Suppose an adversary A intercepts the message send by the Patient’s device ⟨tu,C2,C3⟩ to forge its identity to the application provider. However, in the proposed scheme, an adversary fails to deduce the parameter C1 as it includes a random ephemeral key ymd and nonce nu. Thus, it is difficult for A to obtain two secret parameters to forge successfully. Known session key secrecy In this scheme, the application provider, and patient’s mobile device chooses a secret ephemeral key yap/ymd∈Zp, random nonces na/nu which are generated freshly each time the protocol is run. In the protocol, the session key SK is generated using the combination of nonces, long-term secret key, and ephemeral keys. Therefore, an adversary will fail to re-create the session key with partial information due to the difficulty of solving the ECDLP assumption. ProVerif security analysis In this section, we aim to analyze the proposed CL-AKA protocol using the widely accepted ProVerif tool [45] which is used to verify the security of cryptographic protocols automatically. The tool used pi-calculus language for describing and analyzing protocols. ProVerif supports several cryptographic properties such as encryption/decryption (symmetric and asymmetric), hash functions, and digital signatures. This tool enables session simulation and message space to determines whether the correctness of the protocol can be proved. The adversary is assumed to be able to eavesdrop, insert, and delete the messages. Upon the verification of cryptographic protocol based on required security properties, one of the following may occur:If the proof is true, it states that the attacker is unreachable. This makes ProVerif suitable for proving the secrecy of terms in a protocol. Otherwise, if the proof is false, it states that ProVerif is able to provide an attack trace. Further, it proves security properties like perfect secrecy, mutual authentication, based on which our proposed protocol is verified. Definitions Open channels SecChanl, PubChanl are defined for registration and authentication. The code has few constants like identities IDmd,IDap,IDnm,IDsni and variables (P,xmd,sri, rd,xap,rsni,ra). The operations are string concatenation, XOR operation, hash function, addition, and multiplication. Followed by events that are applied to check correspondence relation in the mutual authentication phase of the proposed scheme. The queries about session keys are to check the secrecy of the key. The definitions are depicted in Fig. 5.Fig. 5 Definition of the code Process The code is written for four entities namely, Patient’s device (MD), AP (Application provider), sensor nodes (SN) and NM (Network manager). The MD, SN and NM processes are depicted in Figs. 6 and 7. It consist of the registration phase of Patient through mobile device, sensor nodes with NM. Whereas, Fig. 8 represents the registration phase of AP with NM and the authentication phase details of exchange of session between AP and Patient’s device for mutual authentication and secure exhange. The detail process of NM’s key generation and work process is also depicted in it.Fig. 6 Process for patient’s device Fig. 7 Process for NM Result The results for the eight queries are demonstrated in Fig. 9. The result of relation query shows that the event(UserNM(a,b)) is correctly executed after the event(UserAuth(a,b)). Similarly, event(APNM(a_17,b_18)) is correctly executed after the event(APAuth(a_17,b_18)), event(begin_SN(a_19)) is correctly executed after the event(SNAuth(a_19)) and inj-event(UA(a_21,b_22,c)) is correctly executed after the inj-event(acceptAU(a_21,b_22,c)). The events are executed in the simulation process RESULTnotattacker(xmd[])istrue, RESULT notattacker(xap[])istrue, and RESULT notattacker(sri[])istrue. This shows that the private keys are secured. Also the RESULT notattacker(SKua[])istrue states that the session keys are secured against various attacks. Thus, the scheme is verified under ProVerif.Fig. 8 Process for AP Fig. 9 Result Performance analysis In this section, the performance analysis of the proposed CL-AKA scheme is discussed in comparison with existing competent schemes namely, [13, 17–19, 29– 31, 33], and [37]. This section demonstrates the comparision of the proposed scheme with respect to security features, computation cost, and communication cost with above mentioned seven related protocols.Table 2 Execution time of various operations Notation Execution time (seconds) Tmul Time complexity for executing the modular multiplication is 0.343 s Texp Time complexity for executing the modular exponentiation is 0.140 s Tsm Time complexity for executing the elliptic curve scalar point multiplication is 0.031 s Tenc/dec Time complexity for executing AES-256 encryption and decryption is 0.937 s Tbp Time complexity for executing the bilinear pairing operation is 4.06 s Th Time complexity for executing the hash function is 0.001 s Comparison of computation cost The evaluation environment is a laptop running Windows 10 and 64-bit Intel(R) Core(TM) i7-10750 H CPU @2.60GHz, 16.00GB RAM. If we consider the schemes based on bilinear pairing, then the Tate pairing e:G1×G1→GT defined on a super-singular curve E1:y2=x3-x+1 mod p where p denotes 160-bit prime number and the size of elements taken for computation in G1 is 320 bits. The state-of-the-art of computing the Tate bilinear pairing is eta pairing, introduced by Barreto et al. [46] is used for implementation. This achieves the security level of the RSA algorithm with a 1024-bit key length. To attain same security level, in the proposed scheme the Koblitz curve secp256k1 defined in Standards for Efficient Cryptography (SEC) [47] is utilized. The curve E2:y2=x3+ax+b mod p where p is 160-bit prime number for a,b∈Zq∗ where q=160 bits and size of elements in G is 320 bits. Table 2, shows the notations for different cryptographic operations along with their execution time in seconds. The computation cost of proposed CL-AKA scheme is compared with existing competent schemes in Table 3.Table 3 Computation cost Schemes Patients AP Total Liu et al. [13] 4Tsm+Texp Tsm+Texp+Tbp 5Tsm+2Texp+Tbp≈4.495 Wang et al. [17] 3Tsm+Tbp 2Tsm+Tbp 5Tsm+2Tbp≈8.275 Wu et al. [18] 3Tsm+4Th+2Texp 3Tsm+4Th+2Texp+Tbp 6Tsm+8Th+4Texp+Tbp≈4.814 He et al. [19] 4Tsm+4Th 4Tsm+2Tbp 8Tsm+4Th+2Tbp≈8.372 Sowjanya et al. [29] 3Tsm+Th+Tmul 6Tsm+3Th+Tmul 9Tsm+4Th+2Tmul≈0.969 Shuai et al. [30] 4Tsm+4Th+Tmul 4Tsm+4Th+Tmul 8Tsm+8Th+2Tmul≈0.942 Kumar et al. [31] 3Tsm+4Th+2Tmul 6Tsm+4Th 9Tsm+8Th+2Tmul≈0.973 Lara  et al. [33] 3Tsm+4Th+Tenc 3Tsm+4Th+Tdec 6Tsm+8Th+2Tenc/dec≈2.069 Cheng  et al. [37] 3Tsm+5Th+2Tmul 5Tsm+3Th+2Tmul 8Tsm+8Th+4Tmul≈1.628 Proposed CL-AKA 4Tsm+3Th+Tmul 4Tsm+4Th+Tmul 8Tsm+7Th+2Tmul≈0.941 Comparison of communication cost In order to compare the communication cost of the proposed CL-AKA scheme with existing ones, let us assume that the length of the identity as |ID| is 32 bits, timestamp |T| is 32 bits, the size of random number |Zq| is 160 bits, the scalar point multiplication as |G| is 320 bits, pairing-based scalar multiplication as |G1| is 320 bits, hash function as |H| is 256 bits, and symmetric enc/dec function ED is 256 bits, respectively. In the proposed scheme the message transferred between Patient to AP contains ⟨tu,C2,C3⟩ which needs (32 + 160 + 160) = 352 bits and response message from AP contains ⟨ta,D2,D3⟩ which needs (32 + 160 + 160) = 352 bits. Therefore, the total communication cost of the proposed CL-AKA scheme is 704 bits. The communication costs of competent existing schemes are depicted in Table 4.Table 4 Communication cost Scheme Communication cost Length (in bits) Liu et al. [13] |T|+2|Zq|+|G1|+2|H| 1184 Wang et al. [17] 2|T|+2|G1|+|ED|+|H| 1216 Wu et al. [18] 2|T|+|Zq|+|ED|+|G1| 800 He et al. [19] |T|+|ED|+2|G1|+|H| 1184 Sowjanya et al. [29] |ED|+2|G|+|H| 1152 Shuai et al. [30] 2|T|+2|G|+3|Zq| 1184 Kumar et al. [31] |T|+2|G|+|Zq|+|H| 1088 Lara  et al. [33] |T|+2|G|+|Zq| 832 Cheng  et al. [37] 3|T|+2|G|+|ED|+3|H| 1760 Proposed CL-AKA 2|T|+4|Zq| 704 Table 5 Security features Scheme [13] [17] [18] [19] [29] [30] [31] [33] [37] Proposed CL-AKA Mutual authentication N Y N Y Y Y Y Y Y Y Resistance against sensor impersonation attack N N N Y Y Y Y Y Y Y Resistance against AP impersonation attack N N N Y Y Y Y Y Y Y Replay attack N Y Y Y Y Y Y Y Y Y Resistance against patient impersonation attack N Y Y Y Y Y Y Y Y Y Known session key secrecy Y Y Y Y Y Y Y Y Y Y Perfect forward secrecy N Y Y Y Y Y N Y Y Y Formal security proof N N Y Y Y Y Y Y N Y Comparison of security and functional features In this subsection, we analyze the security and functional features of the proposed CL-AKA scheme with the existing competent schemes. Table 5 emphasizes on the security features which includes, mutual authentication, impersonation attack, user anonymity, untraceability, session key agreement, perfect forward secrecy, and formal security proof. In this table ’Y’ indicates the security feature is addressed whereas ’N’ indicates the absence of that feature. Conclusion This paper proposes a certificateless authenticated key agreement protocol for remotely monitoring patients health using WBAN. The proposed scheme provides perfect forward secrecy, resistance against sensor/application provider & patient’s device impersonation attack, mutual authentication, and known session key secrecy. The formal security analysis shows that the proposed scheme is able to provide session key security in the widely accepted ROR model. The validation of the proposed CL-AKA scheme using the widely accepted ProVerif tool states that the protocol is safe. In addition, the performance analysis shows that the proposed scheme has low computation and communication cost compared with existing competent schemes. Therefore, the proposed scheme can be applied to e-healthcare applications. Data availibility Not applicable Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References 1. Thotahewa KMS Redouté JM Yuce MR Ultra wideband wireless body area networks 2014 Cham Springer International Publishing 2. Zimmerman TG Personal area networks: Near-field intrabody communication IBM Systems Journal 1996 35 3.4 609 617 10.1147/sj.353.0609 3. Van Dam, K., Pitchers, S., & Barnard, M. (2001). Body area networks: Towards a wearable future. In Proc. WWRF kick off meeting. (pp. 6–7). 4. Sangari, A. S., Manickam, J. M. L. (2014). Public key cryptosystem based security in wireless body area network. In Circuit, power and computing technologies (ICCPCT), 2014 international conference on. (pp. 1609–1612). IEEE. 5. Li J Chen X Li M Li J Lee PPC Lou Wenjing Secure deduplication with efficient and reliable convergent key management IEEE Transactions on Parallel and Distributed Systems 2014 25 6 1615 1625 10.1109/TPDS.2013.284 6. Shen J Zheng W-Y Wang J Zheng Y-H Sun X-M Lee S-Y An efficient verifiably encrypted signature from weil pairing Journal of Internet Technology 2013 14 6 947 952 7. Tan CC Wang H Zhong S Li Q Ibe-lite: A lightweight identity-based cryptography for body sensor networks IEEE Transactions on Information Technology in Biomedicine 2009 13 6 926 932 10.1109/TITB.2009.2033055 19789117 8. Li J Li J Chen X Jia C Lou W Identity-based encryption with outsourced revocation in cloud computing IEEE Transactions on Computers 2015 64 2 425 437 10.1109/TC.2013.208 9. Li X Niu J Liao J Liang Wei Cryptanalysis of a dynamic identity-based remote user authentication scheme with verifiable password update International Journal of Communication Systems 2015 28 2 374 382 10.1002/dac.2676 10. Al-Riyami, S. S. & Paterson, K. G. (2003). Certificateless public key cryptography. In International conference on the theory and application of cryptology and information security (pp. 452–473). Springer. 11. Drira, W., Renault, É. & Zeghlache, D. (2012) A hybrid authentication and key establishment scheme for wban. In 2012 IEEE 11th international conference on trust, security and privacy in computing and communications (pp. 78–83). IEEE. 12. Kompara M Hölbl M Survey on security in intra-body area network communication Ad Hoc Networks 2018 70 23 43 10.1016/j.adhoc.2017.11.006 13. Liu J Zhang Z Chen X Kwak KS Certificateless remote anonymous authentication schemes for wirelessbody area networks IEEE Transactions on Parallel and Distributed Systems 2013 25 2 332 342 10.1109/TPDS.2013.145 14. Liu J Zhang Z Chen X Kwak KS Certificateless remote anonymous authentication schemes for wirelessbody area networks IEEE Transactions on Parallel and Distributed Systems 2014 25 2 332 342 10.1109/TPDS.2013.145 15. Xiong H Cost-effective scalable and anonymous certificateless remote authentication protocol IEEE Transactions on Information Forensics and Security 2014 9 12 2327 2339 10.1109/TIFS.2014.2363553 16. Zhao Z An efficient anonymous authentication scheme for wireless body area networks using elliptic curve cryptosystem Journal of Medical Systems 2014 38 2 13 10.1007/s10916-014-0013-5 24481718 17. Wang C Zhang Y New authentication scheme for wireless body area networks using the bilinear pairing Journal of Medical Systems 2015 39 11 136 10.1007/s10916-015-0331-2 26324170 18. Wu L Zhang Y Li L Shen J Efficient and anonymous authentication scheme for wireless body area networks Journal of Medical Systems 2016 40 6 134 10.1007/s10916-016-0491-8 27091755 19. He D Zeadally S Kumar N Lee JH Anonymous authentication for wireless body area networks with provable security IEEE Systems Journal 2016 11 4 2590 2601 10.1109/JSYST.2016.2544805 20. Xiong H Qin Z Revocable and scalable certificateless remote authentication protocol with anonymity for wireless body area networks IEEE Transactions on Information Forensics and Security 2015 10 7 1442 1455 10.1109/TIFS.2015.2414399 21. Liu J Zhang L Sun R 1-RAAP: An efficient 1-round anonymous authentication protocol for wireless body area networks Sensors 2016 16 5 728 10.3390/s16050728 27213384 22. Li X Peng J Kumari S Wu F Karuppiah M Choo KKR An enhanced 1-round authentication protocol for wireless body area networks with user anonymity Computers & Electrical Engineering 2017 61 238 249 10.1016/j.compeleceng.2017.02.011 23. Khan, H., Dowling, B. & Martin, K.M. (2018). Highly efficient privacy-preserving key agreement for wireless body area Networks. In 2018 17th IEEE international conference on trust, security and privacy in computing and communications/12th IEEE international conference on big data science and engineering (TrustCom/BigDataSE) (pp. 1064-1069). IEEE. 24. Hassan A Omala AA Ali M Jin C Li F Identity-based user authenticated key agreement protocol for multi-server environment with anonymity Mobile Networks and Applications 2019 24 3 890 902 10.1007/s11036-018-1145-5 25. Kumar, P., & Liyanage, M. (2020). Efficient and anonymous mutual authentication protocol in multi-access edge computing (MEC) environments (pp. 119–131). IoT Security: Advances in Authentication. 26. Shen J Gui Z Ji S Shen J Tan H Tang Y Cloud-aided lightweight certificateless authentication protocol with anonymity for wireless body area networks Journal of Network and Computer Applications 2018 106 117 123 10.1016/j.jnca.2018.01.003 27. Tao H Bhuiyan MZA Abdalla AN Hassan MM Zain JM Hayajneh T Secured data collection with hardware-based ciphers for IoT-based healthcare IEEE Internet of Things Journal 2018 6 1 410 420 10.1109/JIOT.2018.2854714 28. Kasyoka P Kimwele M Mbandu Angolo S Certificateless pairing-free authentication scheme for wireless body area network in healthcare management system Journal of Medical Engineering & Technology 2020 44 1 12 19 10.1080/03091902.2019.1707890 31939688 29. Sowjanya K Dasgupta M Ray S An elliptic curve cryptography based enhanced anonymous authentication protocol for wearable health monitoring systems International Journal of Information Security 2020 19 1 129 146 10.1007/s10207-019-00464-9 30. Shuai M Liu B Yu N Xiong L Wang C Efficient and privacy-preserving authentication scheme for wireless body area networks Journal of Information Security and Applications 2020 52 102499 10.1016/j.jisa.2020.102499 31. Kumar M Chand S A lightweight cloud-assisted identity-based anonymous authentication and key agreement protocol for secure wireless body area network IEEE Systems Journal 2020 15 2 2779 2786 10.1109/JSYST.2020.2990749 32. Azees M Vijayakumar P Karuppiah M Nayyar A An efficient anonymous authentication and confidentiality preservation schemes for secure communications in wireless body area networks Wireless Networks 2021 27 3 2119 2130 10.1007/s11276-021-02560-y 33. Lara E Aguilar L García JA Lightweight authentication protocol using self-certified public keys for wireless body area networks in health-care applications IEEE Access 2021 9 79196 79213 10.1109/ACCESS.2021.3084135 34. Soni, M. & Singh, D.K. (2021). LAKA: Lightweight authentication and key agreement protocol for internet of things based wireless body area network. Wireless Personal Communications, 1–18. 35. Peng C Luo M Li L Choo KKR He D Efficient certificateless online/offline signature scheme for wireless body area networks IEEE Internet of Things Journal 2021 8 18 14287 14298 10.1109/JIOT.2021.3068364 36. Liu, S., Chen, L., Wang, H., Fu, S. & Shi, L. (2022). O3HSC: Outsourced online/offline hybrid signcryption for wireless body area networks. IEEE Transactions on Network and Service Management. 37. Cheng Q Li Y Shi W Li X A certificateless authentication and key agreement scheme for secure cloud-assisted wireless body area network Mobile Networks and Applications 2022 27 1 346 356 10.1007/s11036-021-01840-3 38. Li C Xu C Efficient anonymous authentication for wireless body area networks IEEE Access 2022 10 80015 80026 10.1109/ACCESS.2022.3180165 39. Hasan K Chowdhury MJM Biswas K Ahmed K Islam MS Usman M A blockchain-based secure data-sharing framework for software defined wireless body area networks Computer Networks 2022 211 109004 10.1016/j.comnet.2022.109004 40. Dolev D Yao A On the security of public key protocols IEEE Transactions on information theory 1983 29 2 198 208 10.1109/TIT.1983.1056650 41. The Transport Layer Security (TLS) Protocol Version 1.3. Retrieved Aug 2022 from, https://www.rfc-editor.org/rfc/rfc8446.html. 42. Insecure Communication. Retrieved Aug 2022 from, https://owasp.org/www-project-mobile-top-10/2016-risks/m3-insecure-communication. 43. Bellare, M., & Rogaway, p. (1993). Random oracles are practical: A paradigm for designing efficient protocols. In Proceedings of the 1st ACM conference on computer and communications security pp. 62–73. 44. Shoup V Sequences of games: A tool for taming complexity in security proofs IACR Cryptology ePrint Archive 2004 2004 332 45. Blanchet, B., Smyth, B., Cheval, V. & Sylvestre, M. (2018). ProVerif 2.00: Automatic cryptographic protocol verifier, user manual and tutorial. Version from, pp. 05-16. 46. Barreto PS Galbraith SD hÉigeartaigh CÓ Efficient pairing computation on supersingular abelian varieties Designs, Codes and Cryptography 2007 42 3 239 271 10.1007/s10623-006-9033-6 47. Brown, D. R. (2010). Sec 2: Recommended elliptic curve domain parameters. Standars for Efficient Cryptography.

0

PMC9754315

NO-CC CODE

2022-12-16 23:26:35

no

Wireless Netw. 2022 Dec 15;:1-18

utf-8

oa_other

==== Front Child Abuse Negl Child Abuse Negl Child Abuse & Neglect 0145-2134 1873-7757 Elsevier Ltd. S0145-2134(21)00126-5 10.1016/j.chiabu.2021.105053 105053 Article COVID-19 and violence against children: A review of early studies Cappa Claudia * Jijon Isabel UNICEF, Data and Analytics Section, 3 UN Plaza, New York, NY 10017, USA ⁎ Corresponding author. 14 4 2021 6 2021 14 4 2021 116 105053105053 19 1 2021 24 3 2021 25 3 2021 © 2021 Elsevier Ltd. All rights reserved. 2021 Elsevier Ltd Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. Background Throughout the course of the COVID-19 pandemic, researchers across the globe have attempted to understand how the health and socioeconomic crisis brought about by the coronavirus is affecting children’s exposure to violence. Since containment measures have disrupted many data collection and research efforts, studies have had to rely on existing data or design new approaches to gathering relevant information. Objective This article reviews the literature that has been produced on children’s exposure to violence during the pandemic to understand emerging patterns and critically appraise methodologies to help inform the design of future studies. The article concludes with recommendations for future research. Participants and Setting The study entailed a search of working papers, technical reports, and journal articles. Methods The search used a combination of search terms to identify relevant articles and reports published between March 1 and December 31, 2020. The sources were assessed according to scope and study design. Results The review identified 48 recent working papers, technical reports, and journal articles on the impact of COVID-19 on violence against children. In terms of scope and methods, the review led to three main findings: 1) Studies have focused on physical or psychological violence at home and less attention has been paid to other forms of violence against children, 2) most studies have relied on administrative records, while other data sources, such as surveys or big data, were less commonly employed, and 3) different definitions and study designs were used to gather data directly, resulting in findings that are hardly generalizable. With respect to children’s experience of violence, the review led to four main findings: 1) Studies found a decrease in police reports and referrals to child protective services, 2) mixed results were found with respect to the number of calls to police or domestic violence helplines, 3) articles showed an increase in child abuse-related injuries treated in hospitals, and 4) surveys reported an increase in family violence. Conclusions This review underscores the persistent challenges affecting the availability and quality of data on violence against children, including the absence of standards for measuring this sensitive issue as well as the limited availability of baseline data. Future research on COVID-19 and violence against children should address some of the gaps identified in this review. Keywords Violence Children COVID-19 Surveys ==== Body pmc1 Introduction The COVID-19 pandemic has disrupted many aspects of children’s lives and may be putting children around the world at a greater risk of violence. Many of the factors associated with such violence have been exacerbated by COVID-19. Violence at home, for instance, is linked to parental stress, financial hardship and poor mental health (Cicchetti & Carlson, 1989; Jewkes, 2002; Zeanah & Humphreys, 2018). These issues may have deepened with the spread of the disease and with its associated economic and social impacts (Ramaswamy & Seshadri, 2020; Xiong et al., 2020). Likewise, children’s increased presence online might be tied to other forms of violence, such as cyberbullying and online abuse (Yang, 2021). School closures and national lockdowns have also meant that teachers and healthcare workers, who usually identify and report instances of child maltreatment (Feng, Huang, & Wang, 2010; Kenny, 2001; Nayda, 2002), are no longer interacting regularly with children. Violence against children is a serious issue. In the short term, it can lead to severe injuries, dangerous coping behaviors, and even death (Hillis, Mercy, & Saul, 2017). In the long run, it can impair children’s health and development, lead to mental health issues, and contribute to unintended pregnancies and communicable diseases (Ramaswamy & Seshadri, 2020). What is more, children who are exposed to violence are more likely to be victims or perpetrators of violence in the future, in turn affecting new generations (Capaldi, Knoble, Shortt, & Kim, 2012; Tharp et al., 2013). It is therefore imperative that we understand how the current health crisis is affecting children’s exposure to violence, since such exposure may trigger wide-ranging and long-lasting impacts. Since containment efforts have disrupted many data collection and research efforts, studies have often had to rely on evidence from previous crises, existing data or design new approaches to gathering relevant information. Studies of past epidemics have documented impacts on the experience and reporting of violence against children, as well as changes in the delivery of violence prevention and response services. Research conducted during Ebola outbreaks in West and Central Africa found increased reporting of physical violence by children and by community members, resulting from heightened parental stress and tension, children’s increased presence at home and sexual exploitation (International Rescue Committee, 2019; United Nations Development Programme, 2015). There is also evidence of widespread disruptions to child welfare structures and community, and child protection responses (Overseas Development Institute, 2015). During the 2017 cholera outbreak in Yemen, children with sick caregivers slept alone outside treatment centers, which led to an increased risk of sexual violence, particularly among girls (The Alliance for Child Protection in Humanitarian Emergencies, 2018). Despite these examples, evidence on the impacts of previous health crises on violence against children is scarce, and mostly obtained from qualitative studies, thus resulting in limited data (Fraser, 2020). This article reviews the early literature that has been produced on children’s exposure to violence during the COVID-19 pandemic to understand emerging patterns and critically appraise methodologies to help inform the design of future studies. This article then discusses the difficulties in studying COVID-19 and violence against children and illustrates ways researchers have found to fill gaps in available data. The article concludes with recommendations for future research. 2 Methods The World Health Organization (WHO) (2020) identifies six types of violence against children: 1) physical maltreatment and neglect, 2) bullying, 3) youth violence, 4) intimate partner violence, 5) sexual violence, and 6) emotional and psychological violence. This article uses these categories to review the available evidence. The following keyword groups were searched in PsycINFO, Sociological Abstracts, and Google Scholar: 1) child, children, 2) adolescent, adolescence, 3) COVID-19, coronavirus, and 4) violence, with the following variations of: 4.1) physical punishment, maltreatment, harsh parenting, 4.2.) bullying, cyberbullying, 4.3) gang, gangs, community violence, 4.4) intimate partner violence, gender-based violence, child marriage, 4.5) sexual abuse, rape, harassment, voyeurism, online exploitation, and 4.6) emotional abuse, psychological abuse, witness violence. The search included published journal articles, studies produced by non-governmental organizations (NGOs) (labeled here as technical reports), and manuscripts submitted to academic journals that can be found in online repositories such as the Social Science Research Network (labeled here as working papers). Working papers that had not undergone the peer-review process at the time of the search were included because this article contends with an ongoing situation and is an analysis of preliminary research. The search was conducted between October 2020 and January 2021 and only articles published between March 2020 and December 2020 were included. PsycINFO returned 30 articles, Sociological Abstracts 109, and Google Scholar 189. Several articles appeared in at least two of these databases. Twenty-four articles were also reviewed through a manual search using the Google search engine. A total of 273 individual articles were screened. All articles not written in English and that did not include data were excluded (i.e., opinion pieces, letters to the editor, commentaries, etc.). Articles focusing on COVID-19’s impact on crime in general (such as burglary, theft, and murder), as well as articles on people’s attitudes towards domestic violence and child abuse were excluded as well. Titles and abstracts were reviewed to determine eligibility. Full-text articles were then obtained and read. Fig. 1 outlines the selection process. Forty-eight articles were found. Thematic inductive analysis was used to identify the main themes and findings that are summarized in the following section (Nowell, Norris, White, & Moules, 2017).Fig. 1 PRISMA model of the article selection process. Fig. 1 3 Findings Forty-eight papers published between March and December 2020 were identified (the main characteristics of these papers are summarized in Annex 1). Twenty-one articles studied COVID-19 and violence against children in the United States and Canada, five in Europe, three in Latin America and the Caribbean, three in sub-Saharan Africa, and three in South Asia. East Asia and Oceania were covered by two articles each and the Middle East and North Africa was only studied by one article. Eight papers studied COVID-19 and violence against children in more than one country or region of the world. Most of these studies reported preliminary findings, although some are part of larger, ongoing projects. Moreover, the majority of them examined how violence against children is changing (25), how children’s access to services is changing (6), or both (15), while 2 articles probed other issues. From an inductive review of these articles, seven main findings emerged. Three of these findings concern the scope of the research, strengths and limitations of different methods used to arrive at the findings, and types of data sources, while four relate to children’s experience of violence during the pandemic. Finding 1. Studies focused mostly on violence within the home, and more specifically on physical maltreatment and neglect, as well as emotional and psychological violence. Studies rarely investigated other forms of violence against children, often because they did not ask children themselves about their experiences. In order to curb the spread of COVID-19, governments restricted people’s movement and instituted stay-at-home policies. By April 2020, one-third of the world’s population was under lockdown (Buchholz, 2020). Therefore, the first finding is unsurprising: most research on COVID-19 and violence against children has focused on violence within the home. Although all studies were interested in understanding whether such violence has increased during the pandemic, a few also examined the specific stressors that can lead to violence. Brown, Doom, Lechuga-Peña, Watamura, and Koppels (2020), for instance, discussed how in the United States, parental anxiety and depression, more than the pandemic itself, correlates with a higher risk of violence. They found that greater parental support and perceived control during the pandemic are associated with lower perceived stress and child abuse potential. Another example is the Beland, Brodeur, Haddad, and Mikola (2020) study in Canada. They found that employment status and work arrangement are not related to family stress or violence, but that having difficulty fulfilling financial obligations or maintaining social ties is. In other words, parents who work from home or are unemployed but not under financial or social duress are less likely to engage in family or domestic violence. Among the studies that focus on family violence, most of them did not specify who the perpetrator of violence is. It is assumed that it is a parent or caregiver, not a sibling or elderly relative. A study on intrafamilial sexual abuse and a study of sibling violence against children with disabilities are the exceptions (Tener et al., 2020; Toseeb, 2020). In terms of types of violence, almost three-quarters of all reviewed articles discussed physical maltreatment and neglect, emotional and psychological violence, or both. Only seven studies focused on sexual violence, usually within the domestic context. Sexual violence is usually studied along with physical, emotional, and psychological violence; only two reports were found that examined COVID-19’s impact on sexual violence, gender-based violence, and/or adolescent intimate partner violence specifically (International Rescue Committee, 2020; Tener et al., 2020). At least eight studies did not differentiate between domestic violence and violence against children (Fig. 2 ).Fig. 2 Number of articles per type of violence against children. Note: 21 articles cover more than one form of violence. Fig. 2 Four studies discussed youth or community violence, with mixed results. McKay, Metzl, and Piemonte (2020) draw from the United States’ Gun Violence Archive and argue that stay-at-home orders have reduced gun violence in public settings or schools (but increased injuries at home). Jones et al. (2020) analyzed interviews with adolescents in Bangladesh, Ethiopia, Jordan and the State of Palestine, and their participants talked about an increase in community violence, as well as an increase in police brutality. Parkes et al. (2020) made similar claims regarding adolescents in Uganda. Boys are more likely to describe an increase in community violence than girls, probably because girls are more confined to their households. These arguments are difficult to compare, given the different types of data used and the different social contexts. While the search revealed dozens of studies of COVID-19 and intimate partner violence among adults, only two studies spoke to the specific situation of adolescents. One report claimed that adolescent girls in Ethiopia are more scared of child marriage and intimate partner violence currently than before the pandemic (Jones et al., 2020). A survey of refugee and displaced girls and women in 15 African countries likewise stated that 73 percent of participants reported an increase in intimate partner violence (International Rescue Committee, 2020). Finally, only two articles on COVID-19 and bullying were found. Babvey et al. (2020) examined bullying on social media. The authors argued that, since March, there has been a significant increase in abusive and hateful content and cyberbullying on Twitter. Jain, Gupta, Satam, and Panda (2020) studied cyberbullying among adolescents and young adults in India. They found an increase in time spent on social media and online gaming, which is associated with an increase in stalking, derogatory comments, leaking pictures and videos online, and harassment. Finding 2. Most studies relied on administrative records, while other data sources, such as surveys or big data, were less commonly employed. Around half of the studies relied on administrative records from police, child protection services, hospitals or helplines. This type of data reflects reporting of incidents of violence and cannot be used to determine whether violence against children has increased or decreased (UNICEF, 2020b). However, some of the studies had, as their objective, to report on the prevalence of victimization and made statements about actual victimization on the basis of such records. Twenty-two studies analyzed data from surveys. Thirteen surveyed parents only, 4 surveyed children and adolescents only, 2 surveyed both parents and children, and 4 interviewed service providers reporting on their perception of what has occurred to children (Fig. 3 ).Fig. 3 Number of articles per type of method used. Fig. 3 Among the six articles that describe findings reported by children and adolescents is a survey of adolescents and young adults in India (Jain et al., 2020). Also included are qualitative interviews with adolescents in Uganda (Parkes et al., 2020) and in Bangladesh, Ethiopia, Jordan and the State of Palestine (Jones et al., 2020), qualitative interviews with girls and women in refugee, displaced and post-conflict settings in Africa (International Rescue Committee, 2020), and one survey in the Netherlands with both parents and adolescents aged 15–18 (Tierolf, Geurts, & Steketee, 2020). Another study conducted in 37 countries surveyed parents and adolescents ages 11–17 (Save the Children, 2020a). Unlike the studies that surveyed parents alone, studies that focused on or included children tended to address a wider variety of forms of violence, not just physical maltreatment and neglect or emotional and psychological violence. Finding 3. Different definitions and study designs were used to gather the data, resulting in findings that are hardly generalizable. While most of the studies reached similar conclusions, the findings are hardly generalizable and comparable, given the significant differences in scope, definitions and study design. Indeed, a vast range of approaches were used to arrive at the findings. In the case of surveys of parents, a few studies involved participants who had children between specific age ranges – between ages 4 and 10, for instance (Lawson, Piel, & Simon, 2020; Takaku & Yokoyama, 2020) – or with specific characteristics, such as children with special educational needs (Toseeb, 2020). The rest only specified that they were interviewing mothers, parents, or caregivers in general with children under the age of 18. Most parent and caregiver surveys examined disciplinary practices at home, such as harsh parenting (spanking or yelling) or child abuse potential (parental distress, rigidity, or parent-child conflict) (11 studies). Some asked about parents’ perception of violence in other households (2 studies). In most surveys, parents provided self-report data – for example, when parents were asked whether they had spanked their child (Lawson et al., 2020). However, in a few cases they reported on what had occurred to their children – for instance, when parents were asked if their child had been hurt by someone else within the household (Toseeb, 2020). Eight studies asked parents about violence directly, for instance asking respondents whether they agreed with the statement “I swore or cursed my child” or “I hit him/her on the bottom with a belt” (Lawson et al., 2020). Six studies, however, did not ask parents about violence but assessed risk factors, such as parental stress or depression, child behavioral problems or prosocial behavior, and family rigidity or conflict. For example, the surveys asked respondents whether they agree with statements such as “My family fights a lot,” “Children should never disobey,” or “A child needs very strict rules” (Brown, Doom, Lechuga-Peña, Watamura, & Koppels, 2020). Finally, two studies did not describe nor did they provide examples of the types of questions they asked (Poonam, Shama, & Tyagi, 2020; Rashid et al., 2020). The sample sizes in these surveys ranged from 51 to more than 20,000 participants. Only two of the surveys had representative samples, a study in Japan (Takaku & Yokoyama, 2020) and a study in Uganda (Mahmud & Riley, 2021). Takaku and Yokoyama (2020) employed random sampling from about 4.8 million people across the nation who had preregistered as potential survey participants. The Uganda study was part of a larger longitudinal project, allowing researchers to compare the experiences of people before and during the pandemic. Researchers in the Netherlands similarly built on a previous study of domestic violence and child abuse to examine changes after the start of COVID-19 and nationwide lockdowns, although these researchers did not have a representative sample (Tierolf et al., 2020). Most of the other studies recruited participants through social media, such as Facebook ads or Amazon’s Mechanical Turk, or through pre-existing networks, and the majority did not report response rates. With the exception of the Uganda and Netherlands studies, most parent surveys could not compare current findings to pre-existing baseline data. They identified an increase in violence against children by asking parents to assess change, for instance asking if they “yelled/screamed at child(ren) more often” or “spanked or caned child(ren) more often” since the start of the pandemic (Chung, Lanier, & Wong, 2020), or asking parents about their actions in the past year and then asking about their actions in the past week (Lawson et al., 2020). Some studies used standardized question banks, such as the Child Abuse Potential Inventory (Brown et al., 2020), the Conflict Tactics Scale, Parent-Child version (Lawson et al., 2020; Tierolf et al., 2020), the Multidimensional Neglectful Behavior Scale Parent Report (Bérubé et al., 2020), or the Child-Parent Relationship Scale (Russell, Hutchison, Tambling, Tomkunas, & Horton, 2020). Others develop their own measurements for harsh parenting (Chung et al., 2020) or domestic violence (Sharma & Tyagi, 2020). Most studies did not report instrument validity or reliability for these new instruments. Most studies also did not define the concepts used to report on violence against children, although all researchers cite the scholarly literature in order to justify their operationalization of each concept. Of the six studies that surveyed or interviewed children, only three included information on their ethical protocols and ethical review process (Parkes et al., 2020; Save the Children, 2020a; Tierolf et al., 2020). These research projects received clearance from the Save the Children U.S. Ethics Review Committee (Save the Children, 2020a), the University College London Institute of Education, the Uganda Virus Research Institute and Uganda National Council for Science and Technology (Parkes et al., 2020), and the Ethical Review Board of Vrije Universiteit Amsterdam (Tierolf et al., 2020). Finding 4. Studies found a decrease in reports and referrals to police and child protective services. Ten of the reviewed studies examined police reports on domestic violence and referrals to ombudsman offices or child protective services. Given that such data are usually compiled over several years, researchers were able to make comparisons and examine changes in patterns. Most of these studies found that there has been no change or a decrease in referrals, reports, and arrests. One study, for instance, compares police reports in 16 major United States cities over the past few years and argues that there has been no difference in the frequency of serious assaults in residences since COVID-19 response measures (Ashby, 2020). Other papers found a decline in police reports of child abuse in Los Angeles (Barboza, Schiamberg, & Pachl, 2020) and Dallas (Piquero et al., 2020), for example. Researchers similarly showed that there are fewer referrals to child protective services in New York (Rapoport, Reisert, Schoeman, & Adesman, 2020), Florida (Baron, Goldstein, & Wallace, 2020), Georgia (Bullinger, Boy et al., 2020), Indiana (Bullinger, Raissian, Feely, & Schneider, 2020), and Mexico (Cabrera-Hernández & Padilla-Romo, 2020). This literature, at first glance, would seem to suggest a decrease in certain forms of violence against children. Researchers, however, caution against this interpretation. As Rapoport et al. (2020) note, educators and health-care professionals are often the ones making abuse referrals. Therefore, stay-at-home measures may not mean a decrease of violence in practice, only a decrease in the people witnessing the effects of that violence. These authors and others call on teachers, social workers, doctors, and nurses to be vigilant, even if only through the online learning or telehealth format. Likewise, some researchers write about the problems with inferring conclusions from aggregate data. Barboza et al. (2020) find a decrease in police reports on violence against children in Los Angeles, but they also identify “hotspots” of child abuse and neglect in places of severe housing burden, school absenteeism, and financial stress. They recommend considering the data by area, not the city level. Similarly, Piquero et al. (2020) found a decrease in the overall number of police reports in Dallas, but a short spike in reports in the two weeks following stay-at-home orders. They want researchers to more closely examine the timing of police reports. And Anderberg, Rainer, and Siuda (2020) showed that, in London, there was a slight increase in police reports of domestic violence, but a significant increase in searches for domestic violence-related terms online. These researchers invite others to consider alternative sources of data. Finding 5. Studies find mixed results in terms of calls to the police and helplines. While reports, referrals, and arrests regarding family violence and violence against children are in decline, another set of articles (seven in total) argues that there has been an increase in 911 calls and calls to family and domestic violence helplines. Authors find an upsurge in 911 calls in several major cities in the United States (Bullinger, Carr et al., 2020; Hsu & Henke, 2020; Sanga & McCrary, 2020; Mohler et al., 2020). One paper specifically shows an increase in calls from certain city blocks. In their study of domestic violence in 14 U.S. metropolitan areas, Leslie and Wilson (2020) argued that, compared to 2019, 2020 saw more calls from blocks without a history of domestic violence and less calls from blocks with this history. Authors also pointed to varying patterns with respect to changes in calls to domestic violence helplines. In Mexico, one study argued that domestic violence calls for legal services have gone down, but domestic violence calls for psychological services have remained the same or increased during certain weeks of the pandemic (Silverio-Murillo, Balmori de la Miyar, & Hoehn-Velasco, 2020). In Argentina, another study concluded that, when compared to trends from the past three years, a national helpline received fewer calls from the police, but more from the victims of violence (Perez-Vincent & Carreras, 2020). A study comparing 48 child helplines spanning 45 countries showed mixed results (Petrowski, Cappa, Pereira, Mason, & Daban, 2020). Petrowski and colleagues found that the overall number of calls to the helplines has increased. However, the calls related to violence against children have only increased in about half the countries analyzed, while they decreased in the rest of the cases. This decrease may be explained by the fact that teachers and other adults who often report cases to helplines and hotlines are no longer in frequent and close contact with children. Lockdowns and living in close quarters with perpetrators may also limit children’s opportunities to safely reach out for help. Furthermore, they may not be aware that these services are still available. Finding 6. Studies of hospital data found an increase in abuse-related injuries. Three of the reviewed articles examined hospital data, which showed an increase in physical intimate partner violence and physical child abuse injuries in the United Kingdom and the United States, compared to the three years prior (Gosangi et al., 2020; Kovler et al., 2020; Sidpra, Abomeli, Hameed, Baker, & Mankad, 2021). By showing an increase in extreme cases, the authors make a case for an increase in violence against women and children overall. Gosangi et al. (2020), for instance, found that while patients are less likely to report domestic or family violence during COVID-19, medical professionals are more likely to treat injuries related to family violence compared to 2017, 2018, and 2019. Finding 7. Surveys report an increase in violence. Fifteen of the studies reviewed are based on surveys with parents and caregivers. These studies make very consistent claims. Around the world (Save the Children, 2020a) and in Bangladesh (Rashid et al., 2020), Canada (Beland et al., 2020; Bérubé et al., 2020), India (Poonam et al., 2020), Japan (Takaku & Yokoyama, 2020), the Netherlands (Tierolf et al., 2020), New Zealand (Overall, 2020), Singapore (Chung et al., 2020), Uganda (Mahmud & Riley, 2021), the United Kingdom (Toseeb, 2020), and the United States (Brown et al., 2020; Lawson et al., 2020; Russell et al., 2020; Ward & Lee, 2020), parents and caregivers reported an increase in violence against children in the home since the start of the pandemic. Most parents and caregivers admit that since March they are more violent than before – for instance, when asked whether since the outbreak of COVID-19 they are “[resorting] to physical punishment too often” (Save the Children, 2020b). Four studies also reported findings from surveys with service providers. For instance, a survey of 87 NGO representatives in 43 countries indicated that most providers perceive an increase in children’s exposure to violence, especially physical and emotional maltreatment within the home, higher rates of community violence outside the home, and a higher risk of witnessing violence (Wilke, Howard, & Pop, 2020). Likewise, a study of staff in refuges for child survivors of domestic violence in Norway showed that staff are particularly worried about vulnerable children during the pandemic, even though they received fewer requests from clients (Øverlien, 2020). As with the parent surveys, these studies capture the change service providers perceive in the communities they assist. 4 Discussion This article has reviewed early studies on the impact of the COVID-19 pandemic on violence against children. Seven findings have emerged, three on the scope and methods of these studies and four on the actual results. The methodological review revealed a focus on certain forms violence (physical and psychological within the family context), while other forms, such as sexual violence, community violence, and adolescent intimate partner violence have been less investigated. Studies were also found to use different definitions of violence against children and inconsistent study designs, and to rely on administrative data sources, while surveys and especially big data were used less often. In terms of results, studies found decreases in police reports and referrals to child protective services regarding violence against children, increases in violence-related injuries, as well as mixed results with respect to calls to police or domestic violence helplines. Surveys reported a perceived or actual increase in family violence, as reported by service providers or parents/caregivers. 4.1 Implications This review underscores the persistent challenges affecting the availability and quality of data on violence against children. This includes the absence of established, internationally agreed standards for measuring and producing statistics on this sensitive issue as well as the limited availability of baseline data on certain forms of violence (Cappa & Petrowski, 2020). These data issues make it difficult to make broader claims about the changing patterns of violence against children during the COVID-19 pandemic. Different methods, definitions and protocols were used to derive findings about changes in children’s experience of violence, which have resulted in studies of varying scope and quality. In some cases, the studies did not report on whether measures were put in place to ensure the rigorous implementation of the study protocols, nor did they clarify if data collection was undertaken with adequate mechanisms to safeguard the protection of the participants. The absence of information on ethical protocols for the surveys that involved children is of particular concern, given that the COVID-19 crisis brought up additional issues in terms of privacy and confidentiality (UNICEF, 2020a). While it is undeniably important to understand violence against children in the domestic context during COVID-19 given stay-at-home restrictions, this focus nonetheless overlooked possible changes in other forms of violence. For example, limited evidence exists on the effect COVID-19 has had on online abuse. As more children study, socialize, and play on the Internet, researchers must be attentive to the possible risks that children face online. This should occur alongside the assessment of how the crisis has possibly reduced the occurrence of other forms of violence. The assessment of changes in the number of calls to helplines, child protective service referrals, police reports, or hospital records is useful because it is indicative of changes in outreach and services utilization. These administrative data, however, cannot reveal the actual prevalence of violence against children (UNICEF, 2020b) and cannot be used to report on changes in children’s experiences. There are several limitations to using online surveys to study violence against children. First, survey studies can rarely be compared to baseline data. Participants are asked about their experiences before and during the COVID-19 pandemic. This subjective understanding – influenced by bias, memory, level of assessment or mood – does not necessarily reflect the incidence of violent episodes. Second, while online surveys allow researchers to investigate COVID-19 and violence against children more quickly, the majority of the studies examined here did not survey representative samples of the population. Many researchers relied on Facebook ads or Amazon’s Mechanical Turk to disseminate their projects. Researchers made an effort to capture different demographic groups, but there is an inherent bias in who chooses to respond to an online survey. Despite the shortcomings of some of the studies and findings, this review shows that researchers are finding innovative solutions in order to address data problems and respond to research questions. Some researchers used creative methodologies to engage children. Haffejee and Levine (2020) invited children in South Africa to draw and write about their experiences during lockdown. Jones and colleagues used a virtual participatory research model, meaning that researchers interacted with children online, individually and in groups (see Melachowska et al., 2020 for a detailed methodological description). Other researchers are also harnessing the power of social media. Xue, Chen, Chen, Hu, and Zhu (2020) examined how conversations about domestic violence and violence against children are changing on Twitter. Babvey et al. (2020) analyzed testimonials on violence on Reddit forums. Finally, Fabri and colleagues used statistical modelling to estimate the possible effects of COVID-19 on children’s experiences of violent discipline using pre-COVID data from large nationally representative surveys in three low- and middle-income countries. The authors developed a model of how the COVID-19 pandemic could affect risk factors for violent discipline. Country-specific multivariable linear models were then used to estimate the association between risk factors and children’s experience of violent discipline under a “high restrictions” pandemic scenario approximating conditions expected during a period of intense response measures, and a “lower restrictions” scenario with easing of COVID-19 restrictions but with sustained economic impacts (Fabbri et al., 2020). 4.2 Limitations The articles reviewed here may not represent all current discussions of COVID-19 and violence against children. The articles skew to North America, probably given the language requirement; this may also reflect the fact that the majority of research on violence against children is conducted in this region. It makes it difficult to generalize results, also in light of the differences in the spread of the virus and the measures taken by governments to contain it. For instance, while China and later France implemented lockdown measures and internal mobility restrictions, Sweden has not imposed a full lockdown, has not restricted mobility, and has not closed schools, gyms, and restaurants, choosing instead to encourage people to exercise self-restraint (Yan, Zhang, Wu, Zhu, & Chen, 2020). Therefore, studies that discuss the pandemic’s impact on violence against children in one context cannot capture the pandemic’s impact in another. More comparative research is needed to ascertain the pandemic’s impact around the world. There may also be articles currently in the review process that are not yet available for analysis. This preliminary review, however, does offer a first look at the research landscape. This will allow scholars, practitioners, and policymakers to recognize initial patterns in children’s experience of violence during COVID-19 and understand some of the methodological challenges and shortcomings that affect current research. This also points to overlooked areas that need to be studied and understood further, and adjustments to methods that can lead to more robust studies. 5 Conclusion The reviewed articles illustrate the opportunities and challenges faced by researchers studying COVID-19 and violence against children. Future research on COVID-19 and violence against children should address the knowledge gaps and methodological shortcomings identified in this review. Representative surveys should be conducted when it is safe to do so and should include standardized measurement tools. Investments in strengthening the quality and availability of administrative data should be prioritized. Appendix A Supplementary data The following is Supplementary data to this article: Appendix A Supplementary material related to this article can be found, in the online version, at doi:https://doi.org/10.1016/j.chiabu.2021.105053. ==== Refs References Anderberg D. Rainer H. Siuda F. Quantifying domestic violence in times of crisis (No. 8593). CESifo Working Paper 2020 Ashby M.P. Initial evidence on the relationship between the coronavirus pandemic and crime in the United States Crime Science 9 2020 1 16 https://link.springer.com/content/pdf/10.1186/s40163-020-00117-6.pdf Babvey P. Capela F. Cappa C. Lipizzi C. Petrowski N. Ramirez-Marquez J. Using social media data for assessing children’s exposure to violence during the COVID-19 pandemic Child Abuse & Neglect 2020 (in press) Barboza G.E. Schiamberg L.B. Pachl L. A spatiotemporal analysis of the impact of COVID-19 on child abuse and neglect in the city of Los Angeles, California Child Abuse & Neglect 2020 (in press) Baron E.J. Goldstein E.G. Wallace C.T. Suffering in silence: How COVID-19 school closures inhibit the reporting of child maltreatment (forthcoming). Available at SSRN 3601399 2020 Beland L.P. Brodeur A. Haddad J. Mikola D. COVID-19, family stress and domestic violence: Remote work, isolation and bargaining power. (No. 571). Global Labor Organization Discussion Paper 2020 Bérubé A. Clément M.È. Lafantaisie V. LeBlanc A. Baron M. Picher G. How societal responses to COVID-19 could contribute to child neglect Child Abuse & Neglect 2020 (in press) Brown S.M. Doom J.R. Lechuga-Peña S. Watamura S.E. Koppels T. Stress and parenting during the global COVID-19 pandemic Child Abuse & Neglect 2020 (in press) Buchholz K. What share of the world population is already on COVID-19 lockdown? Statista.com. Published April 23, 2020. (Accessed October 29, 2020) 2020 Bullinger L. Boy A. Feely M. Messner S. Raissian K. Schneider W. COVID-19 and alleged child maltreatment Available at SSRN 3702704 2020 Bullinger L.R. Carr J.B. Packham A. COVID-19 and crime: Effects of stay-at-home orders on domestic violence. National Bureau of Economic Research Working Paper 27667 2020 Bullinger L. Raissian K. Feely M. Schneider W. The neglected ones: Time at home during COVID-19 and child maltreatment Available at SSRN 3674064 2020 Cabrera-Hernández F. Padilla-Romo M. Hidden violence: How COVID-19 school closures reduced the reporting of child maltreatment 2020 University of Tennessee, Department of Economics Working paper Capaldi D.M. Knoble N.B. Shortt J.W. Kim H.K. A systematic review of risk factors for intimate partner violence Partner Abuse 3 2012 231 280 22754606 Cappa C. Petrowski N. Thirty years after the adoption of the Convention on the Rights of the Child: Progress and challenges in building statistical evidence on violence against children Child Abuse & Neglect 2020 (in press) Chung S.K.G. Lanier P. Wong P. Mediating effects of parental stress on harsh parenting and parent-child relationship during coronavirus (COVID-19) pandemic in Singapore Journal of Family Violence 2020 Cicchetti D. Carlson V. Child maltreatment: Theory and research on the causes and consequences of child abuse and neglect 1989 Cambridge University Press Cambridge https://psycnet.apa.org/record/1989-98324-000 Fabbri C. Bhatia A. Petzold M. Jugder M. Guedes A. Cappa C. Modelling the effect of the COVID-19 pandemic on violent discipline against children Child Abuse & Neglect 2020 (in press) Feng J.Y. Huang T.Y. Wang C.J. Kindergarten teachers’ experience with reporting child abuse in Taiwan Child Abuse & Neglect 34 2 2010 124 128 20153054 Fraser E. Impact of COVID-19 pandemic on violence against women and girls, VAWG helpdesk research report No. 284 2020 VAWG Helpdesk London Gosangi B. Park H. Thomas R. Gujrathi R. Bay C.P. Raja A.S. Exacerbation of physical intimate partner violence during COVID-19 lockdown Radiology 298 1 2020 E38 E45 32787700 Haffejee S. Levine D.T. “When will I be free?” Lessons from COVID-19 for child protection in South Africa Child Abuse & Neglect 2020 (in press) Hillis S.D. Mercy J.A. Saul J.R. The enduring impact of violence against children Psychology, Health & Medicine 22 4 2017 393 405 Hsu L.C. Henke A. The effect of social distancing on police reports of domestic violence Available at SSRN 2020 International Rescue Committee “Everything on her shoulders”: Rapid assessment on gender and violence against women and girls in Beni, DRC 2019 International Rescue Committee New York International Rescue Committee What happened? How the humanitarian response to COVID-19 failed to protect women and girls 2020 International Rescue Committee New York Jain O. Gupta M. Satam S. Panda S. Has the COVID-19 pandemic affected the susceptibility to cyberbullying in India? Computers in Human Behavior Reports 2 2020 100029 Jewkes R. Intimate partner violence: Causes and prevention Lancet 359 9315 2002 1423 1429 11978358 Jones N. Małachowska A. Guglielmi S. Alam F. Abu Hamad B. Alheiwidi S. I have nothing to feed my family…. COVID-19 risk pathways for adolescent girls in low-and middle-income countries. Report 2020 Gender and Adolescence: Global Evidence London Kenny M.C. Child abuse reporting: Teachers’ perceived deterrents Child Abuse & Neglect 25 1 2001 81 92 10.1016/S0145-2134(00)00218-0 11214815 Kovler M.L. Ziegfeld S. Ryan L.M. Goldstein M.A. Gardner R. Garcia A.V. Increased proportion of physical child abuse injuries at a level I pediatric trauma center during the COVID-19 pandemic Child Abuse & Neglect 2020 10.1016/j.chiabu.2020.104756 (in press) Lawson M. Piel M.H. Simon M. Child maltreatment during the COVID-19 pandemic: Consequences of parental job loss on psychological and physical abuse towards children Child Abuse & Neglect 2020 10.1016/j.chiabu.2020.104709 (in press) Leslie E. Wilson R. Sheltering in place and domestic violence: Evidence from calls for service during COVID-19 2020 10.1016/j.jpubeco.2020.104241 Available at SSRN 3600646 Mahmud M. Riley E. Household response to an extreme shock: Evidence on the immediate impact of the COVID-19 lockdown on economic outcomes and well-being in rural Uganda Working paper 2021 University of Exeter Open Research 10.1016/j.worlddev.2020.105318 McKay T. Metzl J. Piemonte J. Effects of statewide coronavirus public health measures and state gun laws on American gun violence 2020 Available at SSRN 3680050 https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3680050 Melachowska A. Jones N. Abu Hamad B. Al Abbadi T. Al Almaireh W. Alheiwidi S. GAGE virtual research toolkit: Qualitative research with young people about their COVID-19 experiences 2020 Gender and Adolescence: Global Evidence London https://scholar.google.com/scholar_lookup?title=GAGE%20virtual%20research%20toolkit%3A%20Qualitative%20research%20with%20young%20people%20about%20their%20COVID-19%20experiences&author=A.%20Melachowska&publication_year=2020 Mohler G. Bertozzi A.L. Carter J. Short M.B. Sledge D. Tita G.E. Impact of social distancing during COVID-19 pandemic on crime in Los Angeles and Indianapolis Journal of Criminal Justice 2020 10.1016/j.jcrimjus.2020.101692 (in press) Nayda R. Influences on registered nurses’ decision‐making in cases of suspected child abuse Child Abuse Review: Journal of the British Association for the Study and Prevention of Child Abuse and Neglect 11 3 2002 168 178 10.1002/car.736 Nowell L.S. Norris J.M. White D.E. Moules N.J. Thematic analysis: Striving to meet the trustworthiness criteria International Journal of Qualitative Methods 16 1 2017 1 13 https://journals.sagepub.com/doi/full/10.1177/1609406917733847 Overall N. Sexist attitudes and family aggression during COVID-19 lockdown Working paper 2020 PsyArXiv Preprints https://psycnet.apa.org/record/2021-28244-001 Øverlien C. The COVID‐19 pandemic and its impact on children in domestic violence refuges Child Abuse Review 2020 (in press) https://onlinelibrary.wiley.com/doi/10.1002/car.2650 Overseas Development Institute Special feature: The Ebola crisis in West Africa, no. 64 2015 ODI London https://odihpn.org/wp-content/uploads/2015/06/he_64.pdf Parkes J. Datzberger S. Howell C. Kasidi J. Kiwanuka T. Knight L. Young people, inequality and violence during the COVID-19 lockdown in Uganda Working paper. SocArXiv papers 2020 https://osf.io/preprints/socarxiv/2p6hx/ Perez-Vincent S. Carreras E. Evidence from a domestic violence hotline in Argentina. Chapter 1 in: Inter-American Development Bank technical note Nº IDB-TN 2020 https://publications.iadb.org/publications/english/document/COVID-19-Lockdowns-and-Domestic-Violence-Evidence-from-Two-Studies-in-Argentina.pdf Petrowski N. Cappa C. Pereira A. Mason H. Daban R.A. Violence against children during COVID-19: Assessing and understanding change in use of helplines Child Abuse & Neglect 2020 10.1016/j.chiabu.2020.104757 (in press) Piquero A.R. Riddell J.R. Bishopp S.A. Narvey C. Reid J.A. Piquero N.L. Staying home, staying safe? A short-term analysis of COVID-19 on Dallas domestic violence American Journal of Criminal Justice 2020 1 35 https://link.springer.com/article/10.1007/s12103-020-09531-7 Poonam S. Sharma K. Tyagi P. Correlates of domestic violence in relation to physical health and perceived stress during lockdown Shodh Sarita 7 27 2020 66 71 https://www.researchgate.net/publication/344073584_Correlates_of_domestic_violence_in_relation_to_physical_health_and_perceived_stress_during_lockdown_sodha_sarita Ramaswamy S. Seshadri S. Children on the brink: Risks for child protection, sexual abuse, and related mental health problems in the COVID-19 pandemic Indian Journal of Psychiatry 62 Suppl 3 2020 S404 https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7659798/ 33227060 Rashid S.F. Aktar B. Farnaz N. Theobald S. Ali S. Alam W. Fault-lines in the public health approach to COVID-19: Recognizing inequities and ground realities of poor residents’ lives in the slums of Dhaka City, Bangladesh Available at SSRN 3608577 2020 https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3608577 Russell B.S. Hutchison M. Tambling R. Tomkunas A.J. Horton A.L. Initial challenges of caregiving during COVID-19: Caregiver burden, mental health, and the parent–child relationship Child Psychiatry and Human Development 51 5 2020 671 682 https://link.springer.com/article/10.1007/s10578-020-01037-x 32749568 Rapoport E. Reisert H. Schoeman E. Adesman A. Reporting of child maltreatment during the SARS-CoV-2 pandemic in New York City from March to May 2020 Child Abuse & Neglect 2020 10.1016/j.chiabu.2020.104719 (in press) Sanga S. McCrary J. The impact of the coronavirus lockdown on domestic violence. Available at SSRN 3612491 2020 https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3612491 Save the Children Protect a generation: The impact of COVID-19 on children’s lives https://resourcecentre.savethechildren.net/library/protect-generation-impact-covid-19-childrens-lives. Published September 10, 2020. (Accessed January 8, 2021) 2020 Save the Children The hidden impact of COVID-19: On children, research design and methods https://resourcecentre.savethechildren.net/library/hidden-impact-covid-19-children-global-research-series. Published September 10, 2020. (Accessed January 11, 2021) 2020 Sharma K. Tyagi P. Correlates of domestic violence in relation to physical health and perceived stress during lockdown 2020 Working paper https://www.researchgate.net/profile/Pooja-Tyagi-5/publication/344073584_Correlates_of_domestic_violence_in_relation_to_physical_health_and_perceived_stress_during_lockdown_sodha_sarita/links/5f510acd299bf13a319c2c0d/Correlates-of-domestic-violence-in-relation-to-physical-health-and-perceived-stress-during-lockdown-sodha-sarita.pdf Sidpra J. Abomeli D. Hameed B. Baker J. Mankad K. Rise in the incidence of abusive head trauma during the COVID-19 pandemic Archives of Disease in Childhood 2021 10.1136/archdischild-2020-319872 (in press) Silverio-Murillo A. Balmori de la Miyar J.R. Hoehn-Velasco L. Families under confinement: COVID-19, domestic violence, and alcohol consumption. Andrew Young School of Policy Studies Research paper series 2020 (in press) https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3688384 Takaku R. Yokoyama I. What school closure left in its wake: Contrasting evidence between parents and children from the first COVID-19 outbreak Available at SSRN 3693484 2020 https://papers.ssrn.com/sol3/Papers.cfm?abstract_id=3693484 Tener D. Marmor A. Katz C. Newman A. Silovsky J.F. Shields J. How does COVID-19 impact intrafamilial child sexual abuse? Comparison analysis of reports by practitioners in Israel and the US Child Abuse & Neglect 2020 104779 10.1016/j.chiabu.2020.104779 Tharp A.T. DeGue S. Valle L.A. Brookmeyer K.A. Massetti G.M. Matjasko J.L. A systematic qualitative review of risk and protective factors for sexual violence perpetration Trauma, Violence & Abuse 14 2 2013 133 167 10.1177/1524838012470031 The Alliance for Child Protection in Humanitarian Emergencies The guidance note on the protection of children during infectious disease outbreak The Alliance for Child Protection in Emergencies 2018 https://alliancecpha.org/en/child-protection-online-library/guidance-note-protection-children-during-infectious-disease Tierolf B. Geurts E. Steketee M. Domestic violence in families in the Netherlands during the coronavirus crisis: A mixed method study Child Abuse & Neglect 2020 104800 10.1016/j.chiabu.2020.104800 Toseeb U. Sibling conflict during COVID-19 in families with special educational needs and disabilities 2020 PsyArXiv Working paper https://psyarxiv.com/7fgcn/ United Nations Children’s Fund Research on violence against children during the COVID-19 pandemic: Guidance to inform ethical data collection and evidence generation 2020 New York New York https://data.unicef.org/resources/research-on-violence-against-children-during-the-covid-19-pandemic-guidance/ United Nations Children’s Fund Strengthening administrative data on violence against children: Challenges and promising practices from a review of country experiences 2020 UNICEF New York https://data.unicef.org/resources/strengthening-administrative-data-on-violence-against-children/ United Nations Development Programme Ebola recovery in Sierra Leone: tackling the rise in sexual and gender-based violence and teenage pregnancy during the Ebola crisis 2015 UNDP New York https://www.undp.org/content/undp/en/home/librarypage/crisis-prevention-and-recovery/recovering-from-the-ebola-crisis---full-report.html Ward K.P. Lee S.J. Mothers’ and fathers’ parenting stress, responsiveness, and child wellbeing among low-income families Children and Youth Services Review 116 2020 10.1016/j.childyouth.2020.105218 (in press) World Health Organization (WHO) Violence against children Published June 2, 2020. (Accessed October 24, 2020) 2020 https://www.who.int/news-room/fact-sheets/detail/violence-against-children Wilke N.G. Howard A.H. Pop D. Data-informed recommendations for services providers working with vulnerable children and families during the COVID-19 pandemic Child Abuse & Neglect 2020 10.1016/j.chiabu.2020.104642 (in press) Xiong J. Lipsitz O. Nasri F. Lui L.M. Gill H. Phan L. Impact of COVID-19 pandemic on mental health in the general population: A systematic review Journal of Affective Disorders 277 2020 55 64 10.1016/j.jad.2020.08.001 32799105 Xue J. Chen J. Chen C. Hu R. Zhu T. Abusers indoors and coronavirus outside: An examination of public discourse about COVID-19 and family violence on Twitter using machine learning 2020 MedRXiv Working paper https://www.medrxiv.org/content/10.1101/2020.08.13.20167452v1 Yan B. Zhang X. Wu L. Zhu H. Chen B. Why do countries respond differently to COVID-19? A comparative study of Sweden, China, France, and Japan The American Review of Public Administration 50 6-7 2020 762 769 10.1177/0275074020942445 Yang F. Coping strategies, cyberbullying behaviors, and depression among Chinese netizens during the COVID-19 pandemic: A web-based nationwide survey Journal of Affective Disorders 281 2021 138 144 10.1016/j.jad.2020.12.023 33321378 Zeanah C.H. Humphreys K.L. Child abuse and neglect Journal of the American Academy of Child and Adolescent Psychiatry 57 9 2018 637 644 10.1016/j.jaac.2018.06.007 30196867

33965215

PMC9754317

NO-CC CODE

2022-12-16 23:26:35

no

Child Abuse Negl. 2021 Jun 14; 116:105053

utf-8

Child Abuse Negl

10.1016/j.chiabu.2021.105053

oa_other

==== Front Child Abuse Negl Child Abuse Negl Child Abuse & Neglect 0145-2134 1873-7757 The Authors. Published by Elsevier Ltd. S0145-2134(20)30552-4 10.1016/j.chiabu.2020.104897 104897 Article Modelling the effect of the COVID-19 pandemic on violent discipline against children Fabbri Camilla a*1 Bhatia Amiya a1 Petzold Max b2 Jugder Munkhbadar c3 Guedes Alessandra d4 Cappa Claudia c3 Devries Karen a1 a London School of Hygiene and Tropical Medicine, United Kingdom b University of Gothenburg, Sweden c UNICEF, USA d UNICEF Office of Research Innocenti, Italy ⁎ Corresponding author. 1 London School of Hygiene and Tropical Medicine, 15-17 Tavistock Place, WC1H9SH, London, United Kingdom. 2 School of Public Health and Community Medicine, Institute of Medicine, University of Gothenburg, Medicinaregatan 18A, 41390 Göteborg, Sweden. 3 Data and Analytics Section, Division of Data, Analytics, Planning and Monitoring, UNICEF, New York, USA. 4 UNICEF Office of Research Innocenti, Florence, Italy. 22 12 2020 6 2021 22 12 2020 116 104897104897 9 10 2020 11 12 2020 13 12 2020 © 2021 The Authors 2021 Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. Background The COVID-19 pandemic could increase violence against children at home. However, collecting empirical data on violence is challenging due to ethical, safety, and data quality concerns. Objective This study estimated the anticipated effect of COVID-19 on violent discipline at home using multivariable predictive regression models. Participants Children aged 1–14 years and household members from the Multiple Indicator Cluster Surveys (MICS) conducted in Nigeria, Mongolia, and Suriname before the COVID-19 pandemic were included. Methods A conceptual model of how the COVID-19 pandemic could affect risk factors for violent discipline was developed. Country specific multivariable linear models were used to estimate the association between selected variables from MICS and a violent discipline score which captured the average combination of violent disciplinary methods used in the home. A review of the literature informed the development of quantitative assumptions about how COVID-19 would impact the selected variables under a “high restrictions” pandemic scenario, approximating conditions expected during a period of intense response measures, and a “lower restrictions” scenario with easing of COVID-19 restrictions but with sustained economic impacts. These assumptions were used to estimate changes in violent discipline scores. Results Under a “high restrictions” scenario there would be a 35%–46% increase in violent discipline scores in Nigeria, Mongolia and Suriname, and under a “lower restrictions” scenario there would be between a 4%–6% increase in violent discipline scores in these countries. Conclusion Policy makers need to plan for increases in violent discipline during successive waves of lockdowns. Keywords Physical violence Corporal punishment Violent discipline Violence against children COVID-19 Pandemic MICS ==== Body pmc1 Introduction Violent discipline by parents and other caregivers at home is one of the most common forms of violence against children (Devries et al., 2018; UNICEF, 2014). Approximately half of the world’s children below the age of 15 are subjected to physical punishment (UNICEF, 2014), and roughly three in four children between the ages of 2 and 4 years are exposed to psychological aggression and physical punishment on a regular basis (UNICEF, 2017). Violent discipline is defined as any physical punishment and/or psychological aggression, including spanking or physically forcing children to do things; use of guilt, humiliation, the withdrawal of love, or emotional manipulation to control children (UNICEF, 2010). Use of violent discipline varies by country, and may be both a normative behavior and originate from feelings of stress, frustration and lack of self-control (UNICEF, 2010). Violent discipline in all its forms, and regardless of the reasons that motivate its use, is a fundamental violation of children’s rights (United Nations, 1989). At the time of writing in October 2020, population-level data on children’s experiences of violence during the COVID-19 pandemic are either absent or very limited. Recent research efforts have focused on evaluating the effects of the pandemic on reporting of cases of violence relying on administrative data (Baron, Goldstein, & Wallace, 2020; Cabrera-Hernández & Padilla-Romo, 2020). Collection of data from children on their experiences of violence during the current pandemic is not recommended under most circumstances, due to ethical, safety and methodological concerns (Berman, 2020). Policymakers and service providers are lacking robust evidence on how COVID-19 response measures and the socioeconomic impacts of the current crisis may affect levels of violence against children. Prior analyses of Multiple Indicator Cluster Surveys (MICS) – one of the few nationally representative and internationally comparable sources of data on violence against children - suggested that across 86 countries an average of 74% of children aged 1–14 years experienced some form of physical punishment and/or psychological aggression in the past month, ranging from 36% in Cuba to 94% in Ghana (UNICEF, 2019). This study is motivated by concerns that the COVID-19 pandemic has potentially increased children’s risk of experiencing violence. The pandemic has led to the disruption of formal and informal child protection systems responsible for identification of and response to cases of violence (Bhatia et al., 2020; The Alliance for Child Protection in Humanitarian Action, 2020; WHO, 2020) and exacerbated many of the known risk factors for violence against children within the household. Poverty, socioeconomic inequalities, economic insecurity and unemployment compromise caregivers’ mental health and their ability to provide for children, increasing risk of child maltreatment (Berger, 2005; Meinck, Cluver, & Boyes, 2015; Raissian & Bullinger, 2017; Sedlak et al., 2010). Intimate partner violence, tensions in the family, poor mental health, and alcohol use are also associated with increased risk of child abuse and violent discipline (Cluver et al., 2020; Stith et al., 2009; Whipple & Webster-Stratton, 1991). Furthermore, caregivers’ psychological status and subjective wellbeing may influence their relationships with children and therefore their disciplinary methods (Brown, Doom, Lechuga-Peña, Watamura, & Koppels, 2020). This paper has three aims: to present the results of multivariable analyses of risk factors for violent discipline using MICS data for Nigeria, Mongolia and Suriname; to propose a framework of how the COVID-19 pandemic could affect risk factors for violent discipline; to estimate, through a modelling approach (Chawanpaiboon et al., 2019; Moller, Petzold, Chou, & Say, 2017), how the severity of violent discipline could change under two hypothetical pandemic scenarios, and to discuss the benefits and challenges of this approach. 2 Methods We followed four steps. One, we developed a conceptual framework that outlines selected household- and child-level risk factors for violent discipline, and illustrates what aspects of the pandemic might directly or indirectly affect such factors, and therefore children’s experiences of violence in the home. We distinguished between factors that would likely be affected by the COVID-19 pandemic, and factors that would remain unaltered but are important predictors of violent discipline. We refined our framework by mapping these risk factors against the variables available in the MICS for three case-study countries that offered comparable data – Nigeria (2016), Mongolia (2018), and Suriname (2018). Two, we developed a multivariable model to estimate the association between risk factors from our conceptual framework and severity of violent child discipline under non-pandemic conditions. Three, we formulated assumptions to quantify the effect of the COVID-19 pandemic on these risk factors under a “higher restrictions” and “lower restrictions” pandemic scenario. Four, using the regression equation from step two, we estimated predicted changes in severity of violent child discipline under the two pandemic scenarios. 2.1 Conceptual framework: how could the COVID-19 pandemic affect violent discipline? We drew on an ecological framework (Maternowska & Fry, 2018) which defines violence as the result of a multitude of interactions at the individual, interpersonal, family, and community level, to identify pathways to violent discipline in the context of the COVID-19. The framework in Fig. 1 illustrates the possible pathways to violent discipline stemming from three common COVID-19 response measures – business closures, social distancing and restrictions to movement, and school closures – as well as from the general fear and insecurity triggered by the spreading of COVID-19 and by changes in the global social and economic context. This was informed by literature on the known risk factors for violent child discipline, including emerging evidence on pathways to violence under pandemic conditions (Bakrania et al., 2020; Peterman et al., 2020). Availability of data in the MICS was also taken into consideration in the development of the framework. We defined violent discipline as the outcome. The risk factors that were likely to be affected by the COVID-19 pandemic appear in blue whereas other factors associated with violent discipline in the literature, but which we assumed would not be affected by COVID-19, appear in grey.Fig. 1 Conceptual framework: The effect of the COVID-19 pandemic on violent discipline. Fig. 1 The current pandemic undoubtedly produced large changes in the global economy with consequences for both income and wealth levels as well as employment at the household level (ILO, 2020a; Lawson, Piel, & Simon, 2020; The World Bank, 2020a). Similarly its potential effects on mental health have been widely acknowledged (United Nations, 2020). Economic insecurity coupled with stay-at-home orders and widespread fear of contagion contributed to increases in levels of stress and anxiety among caregivers (Jia et al., 2020; Salari et al., 2020; Serafini et al., 2020), and may have heightened the risk of conflicts at home and of substance misuse (Biddle, Edwards, Gray, & Sollis, 2020; Clay & Parker, 2020; Sharma & Borah, 2020). Modifications to lifestyles, habits and caring responsibilities induced by COVID-19 containment measures have also affected individuals ‘psychological status and subjective wellbeing (ILO, 2020b; Kola, 2020) with consequences for the risk of harsh parenting (Chung, Lanier, & Wong, 2020). Family structures and household composition may have also been altered due to COVID-19-related changes in migration flows, employment patterns and economic opportunities (Fisher et al., 2020; The World Bank, 2020b). School and business closures, and movement restrictions have radically altered how and where adults and children spend their time, which may affect children’s exposure to violence at home (Bullinger, Raissian, Feely, & Schneider, 2020; Peterman & O’Donnell, 2020). In our framework children’s time is split between schooling (proxied as attendance) and engagement in labor such as household chores and economic activities. Given the lack of evidence from previous and the current pandemic we assumed no effects of COVID-19 on household characteristics such as demographics, education, values and beliefs around violence. For simplicity, we also assumed no changes in risk factors for violence due to COVID-19 associated mortality or hospitalization and therefore did not include these variables in our framework (estimated mortality rate from COVID-19 as at 23 September 2020 was 17.36 per 100,000 in Suriname and 0.56 per 100,000 in Nigeria; no COVID-19 deaths were recorded in Mongolia). Other known risk factors for violent discipline, such as caregivers’ own experiences of violence, availability of social and support networks, and children’s own mental health were not included in the framework because there was no corresponding variable in the MICS datasets. Similarly, due to the unavailability of data in the MICS, the framework intentionally excludes the potential mitigating effects of economic, financial, and social assistance programs implemented in response to the pandemic. 2.2 Analytical approach to modelling the association between risk factors and violent discipline under non-pandemic conditions 2.2.1 Data sources We used MICS data from Nigeria (2016), Mongolia (2018) and Suriname (2018) to estimate mean values (or proportions) and associations between household- and child-level risk factors and violent discipline under non-pandemic conditions. These countries were selected because their respective MICS datasets included all the variables of interest, they were geographically diverse, and differed in the number of COVID-19 cases (The New York Times, 2020) and stringency of the response measures implemented (Hale et al., 2020). Although we selected countries with similar variables, our analyses should not be interpreted as cross-country comparisons but are aimed at showing how this modelling approach can be applied to different datasets. MICS are cross-sectional, nationally representative household surveys which use a multi-stage sampling approach. A dedicated module asks about the use of disciplinary methods by household members. In the case of Nigeria, this module was administered to the household head and asked about discipline methods used with one randomly selected child aged 1–14 years in each household. In Mongolia and Suriname, the child discipline module was administered to the mother, or if the mother was not alive or not living in the same household, to the primary caregiver of every child aged 1–4 years and/or of one randomly selected child aged 5–14 years. In addition, men and women aged 15−25 years in Nigeria and 15−49 years in Mongolia and Suriname were also interviewed and asked about their alcohol use, well-being, and employment (Nigeria only). Data for this study come from the child, household, men’s and women’s questionnaires in each country. 2.2.2 Outcome The primary outcome was caregiver reported use of violent discipline, measured using an adapted version of the Parent-Child Conflict Tactics Scale (PC-CTS). This includes 8 questions on psychological and physical violent disciplinary practices. Caregivers provided information about disciplinary methods used with the child by any member of the household in the past month, it is therefore impossible to know which household members used the reported violent methods. Furthermore, no information was collected on the frequency of these practices in the past month. We used the eight violent discipline items to construct a violent discipline score as a continuous variable. We assigned each of the eight items a score ranging from 5 to 30 points based on the severity of discipline as defined by the Conflict Tactics Scale (Table 1 ). Acts of violent discipline which constituted psychological aggression (name calling and shouting/yelling) were assigned a score of 5 points. Acts of discipline defined as minor assault (shaking if the child was above 2 years of age, spanking, hitting on the bottom, hitting on the arm of legs) were assigned a score of 10 points. Severe assault (shaking if child is under 2 years of age, hitting on the face, head or ears) was assigned a score of 20 points. Finally, very severe assault (beating with an implement) was assigned a score of 30 points. We calculated a total score for each child which ranged from 0 to 110, where 0 corresponded to no use of violent discipline in the past month and 110 to situations where all eight types of psychological and physical violence were used. Although the score used in our analyses can be interpreted as a proxy for the severity of violent discipline, a complete assessment of severity would need to include a measure of frequency, which is not available in the MICS.Table 1 Conflict Tactics Scale Items. Table 1Item Weighting Type of violence (as per the Conflict Tactics Scale) 1 Shook him/her 10 (if > = 2 years) 20 (if <2 years) Minor assault (but severe if child is less than 2 years) 2 Shouted, yelled at or screamed at him/her 5 Psychological aggression 3 Spanked, hit or slapped him/her on the bottom with bare hand 10 Minor assault 4 Hit him/her on the bottom or elsewhere on the body with something like a belt, hairbrush, stick or other hard object 10 Minor assault 5 Called him/her dumb, lazy or another name like that 5 Psychological aggression 6 Hit or slapped him/her on the face, head or ears 20 Severe assault 7 Hit or slapped him/her on the hand, arm or legs 10 Minor assault 8 Beat him/her up with an implement (hit over and over as hard as one could) 30 Very severe assault 2.2.3 Variable selection We defined exposures as the risk factors that we hypothesized would be affected by COVID-19. We drew on data from the men’s and women’s MICS datasets to construct several aggregate and average household-level measures. These included: (1) a measure of youth employment defined as the proportion of household members aged 15−24 years who have a job over the total number of household members between 15−24 years; (2) a measure of subjective wellbeing defined as overall mean happiness; (3) a measure for average household alcohol consumption defined as the number days in which alcohol was consumed in the past month by men and women in the household. The subjective wellbeing (happiness) variable relied on data from household members aged 15−24 years in the Nigeria dataset, but included all women and men aged 15−49 years in households in Mongolia and Suriname. The youth employment variable was the only variable that was not available in all three case-study countries and only available in the Nigeria MICS. At the household level we also included household wealth quintiles (data on household income was not available in the MICS) and two variables that described the household structure: an indicator for the total number of household members and a variable for whether the household head was a woman. At the child level we used three variables to proxy for children’s time use: we included a measure of school attendance defined as the proportion of children who attended school at any point in the past year, and two measures to capture children’s work defined as the number of hours engaged in household chores in the past week and the number of hours engaged in economic activity in the past week. We also included in the model several covariates. These were defined as risk factors for violent discipline that were unlikely to be affected by the pandemic, such as: the child’s sex and age, whether parents were living in the household, education, ethnicity and religion of the household head. We also included measures of attitudes towards violence: attitudes about physical punishment were defined as the percentage of respondents who believe the child needs to be physically punished to be brought up properly. Attitudes towards domestic violence were defined as the average number of “yes” responses to five items which asked respondents whether wife beating was justified (if she: goes out without telling husband, neglects the children, argues with husband, refuses sex with husband, burns the food). Finally, we included urban/rural residence and geographic region. 2.2.4 Statistical analysis First, we used the MICS data to calculate a violent discipline score. Next, we estimated unadjusted bivariate models between exposures, covariates and violent discipline, using linear regression (results not shown). Only observations with data on the violent discipline outcome and covariates were included in bivariate models. These analyses informed the selection of the final list of exposures to be included in our final multivariable models. Finally, we estimated unadjusted and adjusted multivariable linear models with violent discipline as the outcome, that included all theoretically relevant exposures and the covariates above. Only children with data on the outcome and all the covariates were included in our analytical sample and therefore in the fully adjusted models. To measure the association between the identified exposures and the violent discipline outcome we estimated the following OLS regression:Yijv=β0+ β1XCijv+ β1XHjv+β3DCijv+ β4DHjv+λv +εijv +μv where Yijv is the violent discipline outcome for child i in household j in region v. XCijv is an indicator of child i’ s covariates and XHjv is an indicator of household j’s covariates. DCijv includes the child-level exposures and DHjv includes the household level exposures. λv are region fixed effects and εijv and μv are the error terms. We used a linear model because we specified the outcome as a continuous variable and because of its simplicity compared to multiplicative binary models. However, in order to validate our results, we replicated the analyses using probit regression models which specified the violent discipline outcome as a binary variable taking the value of one if the child had experienced any form of violent discipline in the past month (results not shown). This binary outcome specification is consistent with how MICS report violence discipline prevalence, and follows a widely used approach in the literature (Cuartas et al., 2019; UNICEF, 2010). All analyses were weighted to account for the multi-stage sampling design and were conducted using Stata 16, separately for each country. 2.3 How does the COVID-19 pandemic affect risk factors for violent discipline? We formulated assumptions about magnitude and direction of the effect the COVID-19 pandemic on household and child exposures from our conceptual framework under two scenarios. The “high restrictions” scenario describes the potential situation in the immediate aftermath of the pandemic and represents a situation that countries may have experienced during a phase of intense containment measures. The “lower restrictions” scenario refers to a situation when containment measures may have started to ease but the effects of the economic crisis triggered by the pandemic may have started to intensify. To inform these assumptions, we conducted a literature review and also relied on expert opinion. 2.3.1 Literature review We searched for evidence on the relationship between pandemics, humanitarian and economic crises, natural disasters and any of the exposure variables identified in our conceptual framework. We searched Pubmed, Google Scholar, and EconLit with keywords such as “pandemic”, “epidemic”, “crisis”, “disaster”, “covid19”, “quarantine”, and keywords for each intermediate variable. We also searched websites of academic institutions, NGOs and other organisations involved in the current COVID-19 and past epidemic responses, to identify recent unpublished evidence and working papers. We also reviewed literature cited in UNICEF’s rapid review of evidence on the effect of pandemics on child protection (UNICEF, 2020a). 2.3.2 Selection of estimates We selected the most relevant study describing associations between the COVID-19 pandemic and each exposure variable included in our conceptual framework by applying selection criteria in the following order: relevance to crisis settings (considering studies from pandemics most relevant, followed by other crises), study design (cohort studies, followed by cross-sectional studies), representativeness (nationally representative, geographically representative, or other), and relevant geographical setting (studies from the case-study country, followed by regional and global analyses). Where no quantitative evidence was available, we considered qualitative evidence. We prioritized evidence and projections published by international organizations and by national statistical services where possible. When there was more than one relevant study providing an estimate, we discussed amongst the study team and decided on an estimate by consensus. Effect estimates were extracted and stored into a Google form database that generated Table 2 . Further detail on sources of data and explanatory notes for each assumption are provided in Supplementary Table S1.Table 2 Assumptions Formulation. Table 2Definition of variable in MICS High Restrictions Assumptions Lower Restrictions Assumptions NIGERIA School attendance Attended school during current school year 89% decrease in attendance 1.03% drop in enrolment Child labor Hours of economic activity (child) increase of 41% h/w labor increase of 21% h/w labor Hours of household chores (child) increase of 13% h/w household chores increase of 6.5% h/w household chores Employment status Proportion of young people who have a job 17.32% increase in unemployment 45.02% increase in unemployment Mental health (subjective wellbeing) Average overall happiness among household members aged 15−25 years 50.98% decrease in happiness 7.84% decrease in happiness Household wealth Wealth quintiles unchanged 8.81% increase in the number of poor Alcohol use Days alcohol was used in the past month (men) unchanged 0.5 more days of drinking in past month Days alcohol was used in the past month (women) unchanged unchanged Household structure Number of household members 0.1 person increase 0.1 person increase Female headed household unchanged unchanged MONGOLIA School attendance Attended school during current school year 89% decrease in attendance 0.71% drop in enrolment Child labor Hours of economic activity (child) increase of 41% h/w labor increase of 21% h/w labor Hours of household chores (child) increase of 13% h/w household chores increase of 6.5% h/w household chores Employment status Proportion of young people who have a job N/A N/A Mental health (subjective wellbeing) Average overall happiness among household members 50.98% decrease in happiness 7.84% decrease in happiness Household wealth Wealth quintiles unchanged 26.23% increase in the number of poor Alcohol use Days alcohol was used in the past month (men) unchanged 0.5 more drinks in past month Days alcohol was used in the past month (women) unchanged unchanged Household size and composition Number of household members 0.1 person increase 0.1 person increase Female headed household unchanged unchanged SURINAME School attendance Attended school during current school year 89% decrease in attendance 1.20% drop in enrolment Child labor Hours of economic activity (child) increase of 41% h/w labor increase of 21% h/w labor Hours of household chores (child) increase of 13% h/w household chores increase of 6.5% h/w household chores Employment status Proportion of young people who have a job N/A N/A Mental health (subjective wellbeing) Average overall happiness among household members 50.98% decrease in happiness 7.84% decrease in happiness Household wealth Wealth quintiles unchanged 24.47% increase in the number of poor Alcohol use Days alcohol was used in the past month (men) unchanged 0.5 more drinks in past month Days alcohol was used in the past month (women) unchanged unchanged Household size and composition Number of household members 0.1 person increase 0.1 person increase Female headed household unchanged unchanged Notes: - Evidence on school attendance under both scenarios was drawn from UNESCO and UNICEF while we relied on evidence from the Ebola epidemic to make assumptions on children’s time engagement in household chores and income generating activities. - Employment (available only in the Nigeria data) was adjusted based on evidence specific to the COVID-19 pandemic. - Data on subjective wellbeing and alcohol consumption came from a mix of literature from other crisis settings and from the COVID-19 pandemic in other countries. - Changes in wealth were based on poverty forecasts specific to each case-study country. - Increases in household size were informed by discussions among authors based on emerging qualitative evidence from ongoing studies. In the “high restrictions” scenario we assumed that full school closures and movement restrictions would affect children dramatically, with consequences on school attendance and time use. We also assumed important drops in employment (for Nigeria) and happiness, but no changes in wealth distribution or alcohol use. We hypothesized that wealth – relative to income – would be more resistant to shocks and therefore would only be affected over a longer term. We also assumed small changes in household structure linked to internal movements, return migration and conditions of smart working, resulting in crowding. In the “lower restrictions” scenario we assumed re-opening of schools and relied on regional forecasts on risk of drop-out to formulate our assumptions. We assumed continued effects on employment but a partial recovery of subjective wellbeing as individuals adjust to the COVID-19 context. We also assumed that there would be an increase in both poverty (with consequences on the wealth distribution) and alcohol consumption. We also assumed sustained changes to household structures. These assumptions reflect the availability and quality of the evidence at the time of writing. As new data on the effect of COVID-19 on risk factors for violent discipline become available, these assumptions could be updated. 2.3.3 Modelling approach for the pandemic scenarios For each case-study country we applied our “high restrictions” and “lower restrictions” assumptions to the mean (or proportion) of each exposure variable calculated for the sample of children included in the multivariable regression models. For each exposure variable we then multiplied the new mean (or proportion) by the corresponding beta coefficient obtained from the multivariable regression models and computed the predicted violence discipline score under each COVID-19 scenario. 3 Results 3.1 Sample characteristics The final analytical sample included 1,843 children aged 1–14 years in Nigeria, 1,354 in Mongolia, and 679 in Suriname (Table 3 ). Prior to COVID-19, the prevalence of any experience of violent discipline in the past month in the full MICS sample of children aged 1–14 was 84.9% in Nigeria, 49.1% in Mongolia, and 87.3% in Suriname. These estimates confirm findings from previous analyses of MICS (Ministry of Social Affairs and Public Housing, 2019; National Bureau of Statistics, & UNICEF, 2017; National Statistical Office, 2019). The average violent discipline score among children aged 1–14 years was 32.63 in Nigeria, 7.90 in Mongolia, and 19.75 in Suriname and Fig. 2 shows the distribution of the violent discipline score in each country. Supplementary Table S2 includes further detail on the violent discipline scores. The mean violent discipline score in Nigeria suggests that the average child experienced either a mix of psychological and physical violence (mild or severe), or multiple forms of physical violence (mild or severe), or one form of very severe physical violence in the past month. The score for Mongolia suggests that on average children experienced primarily one to two forms of psychological violence or one form of physical violence. In Suriname, on average children were exposed to either multiple forms of psychological violence and one form of physical violence, or multiple forms of physical violence, or one severe form of physical violence in the past month.Table 3 Sample Descriptives. Table 3Sample characteristics Nigeria (n = 1,843) Mongolia (n = 1,354) Suriname (n = 679) % or mean N % or mean N % or mean N Violent discipline score 32.6 1,843 7.9 1,354 19.8 679 Attended school during current school year  No 5.3% 95 3.5% 48 3.2% 30  Yes 94.7% 1,748 96.5% 1306 96.8% 649 Hours of economic activity in the past week (child) 3.4 1,843 0.9 1,354 0.4 679 Hours of household chores in the past week (child) 7.0 1,843 5.7 1,354 1.5 679 Proportion of young people who have a job (out of all young people in the household) 0.3 1,843 N/A N/A N/A N/A Average overall happiness among household members 4.5 1,843 4.2 1,354 4.1 679 Days alcohol was used in the past month (men) 1.4 1,843 1.3 1,354 2.7 679 Days alcohol was used in the past month (women) 0.4 1,843 0.5 1,843 0.9 679 Wealth Quintiles [ref = Poorest]  Poorest 17.4% 267 23.2% 380 20.2% 124  Second 21.1% 348 21.1% 324 26.5% 172  Middle 23.2% 421 15.6% 234 17.8% 136  Fourth 20.3% 398 18.9% 256 20.3% 143  Richest 18.0% 409 21.3% 160 15.2% 104 Number of household members 11.3 1,843 5.0 1,354 6.5 679 Female headed household  No 97.2% 1,745 95.1% 1,283 62.2% 464  Yes 2.9% 98 4.9% 71 37.8% 215 Child's sex  Male 50.1% 937 49.1% 673 49.7% 328  Female 50.0% 906 50.9% 681 50.3% 351 Child's age (years) 9.6 1,843 8.9 1,354 9.3 679 Mother in the household  No 8.3% 208 2.7% 41 10.5% 50  Yes 91.7% 1,635 97.3% 1,313 89.5% 629 Father in the household  No 6.6% 220 6.4% 97 26.2% 156  Yes 93.4% 1,623 93.6% 1,257 73.8% 523 Education of household head  None or non-formal education 40.8% 601 8.8% 126 8.2% 43  Primary 20.2% 416 14.6% 230 27.5% 190  Secondary 23.3% 479 53.0% 735 59.0% 406  Higher 15.7% 347 23.5% 263 5.3% 40 Ethnic group of household head Hausa 57.6% 775 Khalkh 77.8% 1,028 Indigenous/Amerindian 29.5% 139 Igbo 6.2% 218 Kazakh 6.9% 153 Maroon 6.9% 41 Yoruba 7.2% 222 Other 15.3% 173 Creole 11.3% 89 Other ethnic group 29.0% 628 Hindustani 24.7% 188 Javanese 15.3% 129 Mixed ethnicity 10.5% 80 Other 1.7% 13 Religion of household head Christian 27.5% 822 Buddhist 48.3% 642 Christianity 53.6% 343 Islam 71.6% 996 No religion 41.6% 534 Hinduism 19.5% 148 Other 1.0% 25 Muslim or other 10.0% 178 Islam 15.4% 123 Traditional or other religion 4.9% 27 No religion 6.6% 38 Child needs to be physically punished to be brought up properly  No 34.9% 604 76.5% 1,078 77.5% 542  Yes 65.1% 1,239 23.5% 276 22.5% 137 Average domestic violence norms score (out of 5) 0.4 1,843 0.04 1,843 0.03 679 Urban/rural residence  Urban 35.8% 581 62.1% 681 69.0% 403  Rural 64.2% 1,262 37.9% 673 31.0% 276 Geopolitical Zones North central 17.2% 410 Western 15.6% 390 Paramaribo 31.7% 197 North east 24.1% 257 Khangai 18.4% 261 Wanica 32.0% 150 North west 40.7% 569 Central 17.1% 230 Nickerie 5.4% 72 South east 4.1% 171 Eastern 7.8% 185 Coronie 1.0% 11 South south 6.4% 213 Ulaanbaatar 41.2% 288 Saramacca 3.2% 54 South west 7.6% 223 Commewijne 5.8% 65 Marowijne 5.0% 45 Para 7.3% 46 Brokopondo 5.0% 19 Sipaliwini 3.8% 20 Fig. 2 Distribution of violent discipline scores in Nigeria, Mongolia and Suriname. Fig. 2 The mean age of children in the analytical sample was 9 years in each country and the majority of children in each country had attended school in the past year. In Nigeria and Mongolia children spent between 6 and 7 h doing household chores in the past week compared to 1.5 h in Suriname. In Nigeria children spent an average of 3.4 h engaging in economic activities compared to 0.9 h in Mongolia and 0.4 h in Suriname. Sixty-five percent of household heads agreed that a child needs to be physically punished to be brought up properly in Nigeria compared to less than 25% of mothers/primary caregivers in Mongolia and Suriname. Overall, there were no large differences between the analytical sample and the full sample for each MICS country (Supplementary Table S3). 3.2 Base multivariable violent discipline model Table 4 shows null and fully adjusted models for each country. In Nigeria, attitudes supportive of physical punishment were a statistically significant predictor of increased violent discipline. Higher levels of happiness among young people in the household were associated with a lower violent discipline score. In Mongolia, children’s sex (female) and older age appeared to be negatively associated with violent discipline, whereas attitudes supportive of physical punishment were positively associated with violent discipline. Higher levels of average household happiness were weakly associated with a lower violent discipline score in Mongolia (p = 0.093). In Suriname, attitudes in support of physical punishment were a significant risk factor for increased violent discipline score. Reports of violent discipline from any household member were also positively associated with the presence of the mother in the household. Finally, child’s sex (female) and older age, together with higher levels of average household happiness were statistically significant protective factors in Suriname.Table 4 Multilevel Linear Models: Association Between Risk Factors And Violent Discipline Score. Table 4 Nigeria Mongolia Suriname Coef. P 95% Confidence Interval Coef. P 95% Confidence Interval Coef. P 95% Confidence Interval Null model (n = 21,583) Null model (n = 11,131) Null model (n = 6,620) Constant 29.78 0.000 28.97 30.60 7.49 0.000 7.09 7.89 19.62 0.000 18.71 20.54 Analytic sample, null model (n = 1,843) Analytic sample, null model (n = 1,354) Analytic sample, null model (n = 679) Constant 32.63 0.00 30.18 35.08 7.90 0.00 6.72 9.08 19.75 0.00 17.83 21.67 Fully adjusted model (n = 1,843) Fully adjusted model (n = 1,354) Fully adjusted model (n = 679) Constant 37.00 0.001 15.98 58.02 11.84 0.044 0.30 23.38 35.74 0.000 18.13 53.36 Attended school during current school year (2013−2014) [Ref = No] −0.51 0.916 −9.92 8.91 1.36 0.617 −3.99 6.71 −2.39 0.544 −10.14 5.36 Hours of economic activity in the past week (child) 0.08 0.550 −0.19 0.35 0.23 0.044 0.01 0.44 0.21 0.424 −0.30 0.72 Hours of household chores in the past week (child) 0.07 0.398 −0.10 0.24 −0.01 0.855 −0.08 0.07 −0.38 0.096 −0.82 0.07 Proportion of young people who have a job (out of all young people in the household) 2.40 0.327 −2.40 7.20 N/A N/A N/A N/A N/A N/A N/A N/A Average overall happiness among household members aged 15−25 years −4.69 0.006 −8.04 −1.34 −2.10 0.093 −4.54 0.35 −3.32 0.018 −6.06 −0.57 Wealth Quintiles [ref = Poorest] Second −1.23 0.675 −7.00 4.54 2.38 0.159 −0.93 5.69 0.00 1.000 −5.04 5.05 Middle 0.01 0.996 −5.96 5.99 −0.28 0.843 −3.06 2.50 2.93 0.350 −3.23 9.10 Fourth 0.58 0.848 −5.37 6.53 0.02 0.988 −3.31 3.36 1.48 0.588 −3.88 6.84 Richest −2.16 0.588 −10.01 5.68 0.03 0.989 −4.27 4.33 −0.66 0.877 −8.99 7.67 Days alcohol was used in the past month (men) −0.03 0.790 −0.29 0.22 −0.26 0.255 −0.72 0.19 −0.01 0.966 −0.32 0.30 Days alcohol was used in the past month (women) 0.04 0.894 −0.49 0.56 0.12 0.696 −0.49 0.74 −0.10 0.658 −0.52 0.33 Number of household members 0.01 0.955 −0.47 0.50 −0.01 0.976 −0.67 0.65 0.29 0.366 −0.34 0.92 Female headed household [ref = No] 14.20 0.088 −2.09 30.49 0.30 0.871 −3.38 3.99 −1.58 0.351 −4.92 1.75 Child's sex [ref = Male] Female −1.38 0.418 −4.72 1.96 −2.97 0.001 −4.79 −1.16 −4.02 0.010 −7.09 −0.95 Child's age −0.35 0.291 −1.00 0.30 −0.41 0.011 −0.72 −0.09 −0.46 0.098 −1.01 0.09 Mother in the household [ref = No] 3.29 0.299 −2.92 9.50 3.74 0.199 −1.97 9.44 5.20 0.019 0.86 9.53 Father in the household [ref = No] 6.20 0.205 −3.39 15.79 1.76 0.480 −3.13 6.65 3.03 0.118 −0.77 6.84 Education of household head [ref = none or non-formal education] Primary −4.00 0.096 −8.72 0.71 −2.00 0.380 −6.48 2.48 −1.50 0.688 −8.82 5.83 Secondary 0.17 0.949 −5.20 5.55 −1.10 0.618 −5.41 3.22 0.22 0.954 −7.42 7.87 Higher −2.60 0.421 −8.95 3.74 −0.08 0.971 −4.70 4.53 −4.42 0.389 −14.50 5.66 Ethnic group of household head [ref=Hausa] [ref = Khalkh] [ref = Indigenous/Amerindian] Igbo 1.44 0.756 −7.67 10.55 Kazakh −2.13 0.274 −5.95 1.69 Maroon −6.53 0.031 −12.44 −0.61 Yoruba 9.02 0.147 −3.17 21.21 Other 1.75 0.307 −1.61 5.11 Creole −3.06 0.335 −9.31 3.18 Other ethnic group 1.42 0.605 −3.96 6.80 Hindustani −7.84 0.048 −15.62 −0.06 Javanese −8.37 0.036 −16.20 −0.54 Mixed ethnicity −10.80 0.000 −16.20 −5.41 Other −12.08 0.049 −24.11 −0.05 Religion of household head [ref = Christian] [ref = Buddhist] [ref = Christian] Islam 1.32 0.636 −4.14 6.77 No religion −2.22 0.049 −4.42 −0.01 Hinduism 3.196 0.354 −3.584 9.976 Other −4.66 0.361 −14.66 5.35 Muslim and other 3.81 0.046 0.07 7.55 Islam 5.61 0.103 −1.14 12.35 Traditional or other religion −1.81 0.637 −9.35 5.72 No religion −1.71 0.640 −8.89 5.48 Child needs to be physically punished to be brought up properly [ref = No] 20.85 0.000 17.19 24.51 5.15 0.000 2.28 8.01 11.102 0 6.9076 15.295 Average domestic violence norms score 4.15 0.077 −0.44 8.74 −0.90 0.807 −8.11 6.32 15.84 0.213 −9.15 40.83 Urban/rural residence [ref = urban] Rural −0.41 0.869 −5.23 4.42 0.50 0.666 −1.79 2.80 −1.17 0.707 −7.26 4.93 Regions/Geopolitical Zones [ref = North central] [ref = Western] [ref = Paramaribo] North east −10.53 0.001 −16.96 −4.09 Khangai 2.48 0.107 −0.54 5.51 Wanica 1.52 0.515 −3.07 6.10 North west −2.32 0.490 −8.90 4.26 Central 2.61 0.124 −0.72 5.94 Nickerie −5.96 0.021 −11.03 −0.90 South east −0.62 0.902 −10.55 9.31 Eastern 2.96 0.093 −0.49 6.42 Coronie −3.54 0.397 −11.74 4.66 South south −3.95 0.220 −10.27 2.37 Ulaanbaatar 4.48 0.022 0.66 8.30 Saramacca 0.70 0.872 −7.84 9.24 South west −12.65 0.013 −22.58 −2.72 Commewijne −3.88 0.276 −10.87 3.11 Marowijne 1.96 0.640 −6.28 10.20 Para 0.03 0.994 −7.54 7.60 Brokopondo −4.79 0.378 −15.48 5.89 Sipaliwini 9.42 0.163 −3.84 22.67 3.3 Pandemic scenarios Tables 5a, 5b and 5c describe associations between the exposures included in our framework and violent discipline under our base scenario, “high restrictions” and “lower restrictions” pandemic scenarios. Although we predicted similar percentage increases in violent discipline across countries, the size of these increases should be interpreted in reference to the base estimate of violent discipline in each country.Table 5a Modelling The Effects of COVID-19 on Violent Discipline Score In Nigeria. Table 5aExposures Base model Scenario 1: High COVID-19 restrictions Scenario 2: Lower COVID-19 restrictions (A) Proportion or mean (B) Multivariate regression coefficient (C) Violent discipline score for "average child" [A*B] Assumption (E) Proportion or mean (F) Violent discipline score for "average child" [B*E] Assumption (G) Proportion or mean (H) Violent discipline score for "average child" [B*G] Model constant Constant 37.00 37.00 N/A N/A 37.00 N/A N/A 37.00 Intermediate variables Attended school during current school year (child) 0.95 −0.51 −0.48 89% decrease 0.10 −0.05 1.03% decrease 0.94 −0.47 Hours of economic activity in the past week (child) 3.41 0.08 0.28 41% increase 4.81 0.39 21% increase 4.13 0.34 Hours of household chores in the past week (child) 7.05 0.07 0.52 13% increase 7.96 0.58 6.5% increase 7.51 0.55 Proportion of young people who have a job 0.33 2.40 0.78 17.32% decrease 0.27 0.65 45.02% decrease 0.18 0.43 Average happiness among household members aged 15−25 years 4.54 −4.69 −21.31 50.98% decrease 2.23 −10.45 7.84% decrease 4.19 −19.64 Wealth Quintiles Poorest 0.17 0.00 No change 0.17 0.00 8.81% increase in poor 0.26 0.00 Second 0.21 −1.23 −0.26 0.21 −0.26 0.19 −0.23 Middle 0.23 0.01 0.00 0.23 0.00 0.21 0.00 Fourth 0.20 0.58 0.12 0.20 0.12 0.18 0.10 Richest 0.18 −2.16 −0.39 0.18 −0.39 0.16 −0.34 Days alcohol was used in the past month (men) 1.43 −0.03 −0.05 No change 1.43 −0.05 0.5 days increase 1.93 −0.07 Days alcohol was used in the past month (women) 0.36 0.04 0.01 No change 0.36 0.01 No change 0.36 0.01 Number of household members 11.30 0.01 0.16 0.1 person increase 11.40 0.16 0.1 person increase 11.40 0.16 Female headed household 0.03 14.20 0.40 No change 0.03 0.40 No change 0.03 0.40 Other covariates Sum of all covariates 15.85 15.85 15.85 Violent discipline score 32.63 43.97 34.09 Change in violent discipline score 34.75% 4.48% Table 5b Modelling The Effects of COVID-19 on Violent Discipline Score In Mongolia. Table 5bExposures Base model Scenario 1: High COVID-19 restrictions Scenario 2: Lower COVID-19 restrictions (A) Proportion or mean (B) Multivariate regression coefficient (C) Violent discipline score for "average child" [A*B] Assumption (E) Proportion or mean (F) Violent discipline score for "average child" [B*E] Assumption (G) Proportion or mean (H) Violent discipline score for "average child" [B*G] Model constant Constant 11.84 11.84 N/A N/A 11.84 N/A N/A 11.84 Intermediate variables Attended school during current school year (child) 0.97 1.36 1.32 89% decrease 0.11 0.14 0.71% decrease 0.96 1.31 Hours of economic activity in the past week (child) 0.92 0.23 0.21 41% increase 1.30 0.29 21% increase 1.11 0.25 Hours of household chores in the past week (child) 5.66 −0.01 −0.04 13% increase 6.39 −0.04 6.5% increase 6.02 −0.04 Proportion of young people who have a job N/A N/A N/A N/A N/A N/A N/A N/A N/A Average happiness among household members 4.18 −2.10 −8.75 50.98% decrease 2.05 −4.29 7.84% decrease 3.85 −8.06 Wealth Quintiles Poorest 0.23 No change 0.23 0.00 26.23% increase in poor 0.49 0.00 Second 0.21 2.38 0.50 0.21 0.50 0.14 0.34 Middle 0.16 −0.28 −0.04 0.16 −0.04 0.09 −0.03 Fourth 0.19 0.02 0.00 0.19 0.00 0.12 0.00 Richest 0.21 0.03 0.01 0.21 0.01 0.15 0.00 Days alcohol was used in the past month (men) 1.31 −0.26 −0.35 No change 1.31 −0.35 0.5 additional days 1.81 −0.48 Days alcohol was used in the past month (women) 0.49 0.12 0.06 No change 0.49 0.06 No change 0.49 0.06 Number of household members 5.04 −0.01 −0.05 0.1 person increase 5.14 −0.05 0.1 person increase 5.14 −0.05 Female headed household 0.05 0.30 0.01 No change 0.05 0.01 No change 0.05 0.01 Other covariates Sum of all covariates 3.18 3.18 3.18 Violent discipline score 7.90 11.27 8.34 Change in violent discipline score 42.64% 5.61% Table 5c Modelling The Effects of COVID-19 on Violent Discipline Score In Suriname. Table 5cExposures Base model Scenario 1: High COVID-19 restrictions Scenario 2: Lower COVID-19 restrictions (A) Proportion or mean (B) Multivariate regression coefficient (C) Violent discipline score for "average child" [A*B] Assumption (E) Proportion or mean (F) Violent discipline score for "average child" [B*E] Assumption (G) Proportion or mean (H) Violent discipline score for "average child" [B*G] Model constant Constant 35.74 35.74 N/A N/A 35.74 N/A N/A 35.74 Intermediate variables Attended school during current school year (child) 0.97 −2.39 −2.31 89% decrease 0.11 −0.25 1.20% decrease 0.96 −2.29 Hours of economic activity in the past week (child) 0.42 0.21 0.09 41% increase 0.59 0.12 21% increase 0.51 0.11 Hours of household chores in the past week (child) 1.52 −0.38 −0.57 13% increase 1.72 −0.65 6.5% increase 1.62 −0.61 Proportion of young people who have a job N/A N/A N/A N/A N/A N/A N/A N/A N/A Average overall happiness among household members 4.15 −3.32 −13.75 50.98% decrease 2.03 −6.74 7.84% decrease 3.82 −12.67 Wealth Quintiles Poorest 0.20 0.00 No change 0.20 0.00 24.47% decrease 0.45 0.00 Second 0.26 0.00 0.00 0.26 0.00 0.20 0.00 Middle 0.18 2.93 0.52 0.18 0.52 0.12 0.34 Fourth 0.20 1.48 0.30 0.20 0.30 0.14 0.21 Richest 0.15 −0.66 −0.10 0.15 −0.10 0.09 −0.06 Days alcohol was used in the past month (men) 2.73 −0.01 −0.02 No change 2.73 −0.02 0.5 additional days 3.23 −0.02 Days alcohol was used in the past month (women) 0.94 −0.10 −0.09 No change 0.94 −0.09 No change 0.94 −0.09 Number of household members 6.50 0.29 1.89 0.1 person increase 6.60 1.92 0.1 person increase 6.60 1.92 Female headed household 0.38 −1.58 −0.60 No change 0.38 −0.60 No change 0.38 −0.60 Other covariates Sum of all covariates −1.35 −1.35 −1.35 Violent discipline score 19.75 28.81 20.63 Change in violent discipline score 45.86% 4.47% In Nigeria, we estimated that the violent discipline score would change from 32.63 prior to COVID-19 to 43.97 in the “high restrictions” scenario, representing a 34.75% increase. In the “lower restrictions” scenario we predicted a violent discipline score for Nigeria of 34.09, which corresponds to a 4.48% increase from the non-pandemic score. In the context of Nigeria, where the initial violent discipline score was higher, these estimated increases under pandemic conditions mean that on average children could be exposed to repeated forms of physical violence or to forms of very severe beating. In Mongolia, we estimated that the violent discipline score would increase from 7.90 to 11.27, in the “high restrictions” scenario representing a 42.64% increase. In the “lower restrictions” scenario we estimated a violent discipline score of 8.34 which corresponds to a 5.61% increase compared to the base level. Mongolia is one of the countries with the lowest prevalence of violent discipline globally (UNICEF, 2019). Our findings show that although children may become increasingly exposed to violent discipline during periods of high COVID-19 restrictions the severity and types of violence under COVID-19 restrictions remain relatively lower than in other contexts. In Suriname, we estimated an increase in the average violent discipline score from 19.75 to 28.81 in the “high restrictions” scenario, representing a 45.86% increase. In the “lower restrictions” scenario we predicted a violent discipline score of 20.63 which corresponds to a 4.47% increase from our base model. In periods of high restrictions, children could be exposed to either a mix of psychological and physical violence, or multiple forms of physical violence, or one form of very severe physical violence in the past month. The sensitivity analyses conducted with the outcome constructed as a binary variable (results not shown) confirmed the same patterns estimated with the linear modelling approach. Although the violent discipline score prior to COVID-19 varied by country, these findings suggest that, on average, children may be exposed to more violent discipline as a consequence of COVID-19 measures. 4 Discussion 4.1 Summary of main findings Our findings indicate that the COVID-19 pandemic is likely to affect children’s experiences of violent discipline at home. There were large differences in the violent discipline score prior to COVID-19 in each country, which is essential in informing the interpretation of the results from the multivariable models. Under a “high restrictions” COVID-19 scenario we estimate a 35–46% increase in violent discipline scores from their respective base levels in each country. Modelling the longer-term “lower restrictions” scenario, that assumed some easing of restrictions combined with sustained economic effects, suggests a 4–6% increase in violent discipline scores. Our analyses also indicate that reductions in levels of happiness among household members could be a key driver of increases in violent discipline. These results should not be interpreted as changes in the proportion or the number of children who have experienced violent discipline as a result of COVID-19, nor are intended to provide cross-country comparisons. Taken together, they point to increases in the severity of household violence during successive waves of lockdowns. Results suggest that violence prevention should be central to COVID-19 response measures. 4.2 Strengths and limitations The approach we used has both strengths and limitations. We used microdata from MICS surveys, which are nationally representative, internationally comparable and available for over 100 countries. The vast majority of these surveys have relevant data on a range of indicators which are affected by COVID-19 and which are associated with violent discipline. Data on violent discipline were collected using the PC-CTS in all settings, which measures whether specific behavioral acts were used against children in households. We acknowledge that whether survey participants define these acts as ‘violence’ is likely to differ across countries, which has important implications for how results are interpreted. However, our results show that, regardless of differences in interpretation of what constitutes violence, the COVID-19 pandemic is likely to make the average combination of behavioral disciplinary acts that children experience at home more severe. We found that we were able to operationalize most of the major pathways by which COVID-19 pandemic may have affected violent discipline. This means that we were able to fit a statistical model with robust individual level data to describe the associations between household, child level exposures, relevant covariates, and our violent discipline outcome. We formulated assumptions on how COVID-19 may affect household and child level indicators but relied on survey data to model our base scenario. The multivariable regression modelling approach we have taken is replicable in other settings and accessible to a range of professionals with adequate training in statistics, and does not require specialist knowledge of mathematical modelling. This analytical approach can be adapted to other datasets and settings, and as new data become available these models could also be updated to reflect emerging evidence. The limitations of this approach relate mainly to common limitations of modelling. Our approach assumes independence in how COVID-19 affected each individual household and child level exposure, which may have led to overestimation of some of the effects. Although MICS surveys produce a wealth of robust data, we combined data from the men’s, women’s household, and child datasets and therefore derived our regression equation from an analytic sub-sample of households with 1–14 year-old children which also had data on resident 15−24-year old in Nigeria and 15−49 year old in Suriname and Mongolia. MICS surveys do not collect income data for household members, and information on employment was only available in the Nigeria survey, among 15−24 year old, and was measured with only one question. This means that the economic effects of COVID-19 on households are likely to be underestimated. Children’s time use is not directly captured in the MICS, so we estimated this using proxy variables, which could have biased our estimates upwards or downwards. For simplicity, we assumed that children’s time was split between schooling and labor within or outside the household only. School attendance may not adequately capture the longer-term effects of school closures - for example, emerging qualitative data from lockdown suggests that poor young people in Uganda are extremely concerned that they will not be able to return to school and will instead need to spend more time earning income to support their families (Parkes et al., 2020). In some countries, COVID-19 has had large impacts on emotional distress. In the MICS data, we made use of an indicator for ‘overall happiness’ (only measured among 15−24 year old in Nigeria), to capture some variation in wellbeing and mental health, but this is not a validated measure for mental health. If different variables to proxy other mental health impacts had been available, our estimates may have changed up or down. Given that MICS surveys only collect data on violent discipline, our analyses are unable to provide insights into the effects of the pandemic on other forms of violence against children. We were also limited by the data available to inform the assumptions about how COVID-19 affects household and child exposures in our model. For some variables we could rely on official projections and recent estimates from Nigeria, Mongolia and Suriname, however for other variables we had to use less reliable sources of data extrapolating from literature from past health and economic crises and/or from COVID-19-related evidence generated in very different contexts. This means that the final estimates produced for each country under different scenarios reflect only the assumptions outlined in Table 1. Estimates do not fully reflect the nuances of all restrictions in each country and do not account for mitigation measures. Finally, our modelling approach does not allow us to estimate confidence intervals or other measures of variance or precision of our estimates. We are therefore unable to determine the statistical significance of the predicted changes in violent discipline scores. In light of these limitations, our study primarily aimed to illustrate a methodological approach to estimate predicted effects of the pandemic on violent discipline in the absence of current population-level data, and in light of current challenges of collecting data on children’s experiences of violence (UNICEF, 2020b). 4.3 Comparison to other literature There have been a range of efforts to estimate the effects of COVID-19 on various forms of violence; however a recent review of the evidence on COVID-19 and violence against women and children found only three studies that attempted to measure the effects of the pandemic on children’s experiences of violence (Peterman & O’Donnell, 2020). All three studies were conducted in middle to high-income settings and relied on administrative data to assess the effects of the pandemic on reporting of cases of abuse. Sidpra, Abomeli, Hameed, Baker and Mankad (2020) used data from the Hospital for Children NHS Foundation Trust in London to estimate an increase of 1493% in cases of abusive head trauma in the period between March and April 2020 compared to the 3 previous years (Sidpra et al., 2020). Using child maltreatment reports from Indiana’s Child Protective Services between January and May 2020, Bullinger et al. (2020) found a decline in reported cases in April and May 2020 (Bullinger et al., 2020). Similarly, using child maltreatment case reports from the Mexico City Attorney General’s Office, Cabrera-Hernández and Padilla-Romo (2020) used quasi-experimental methods to estimate the effects of school-closures on detection and reporting of cases and found a 21–30% reduction with larger effects among girls and in poorer municipalities (Cabrera-Hernández & Padilla-Romo, 2020). An analysis of data from a survey of 48 child helplines revealed that the number of contacts to helplines has drastically increased during the COVID-19 pandemic and that the number of contacts related to cases of violence has increased in some countries, whereas it decreased in others (Petrowski, Cappa, Pereira, Mason, & Daban, 2020). Importantly, analyses of reports of violence may not be reflective of changes in prevalence given the existing barriers to reporting and help seeking, particularly in the context of COVID-19. Only one study used a similar modelling approach to estimate the effects of COVID-19 on sexual and reproductive health outcomes relying on survey data (Riley, Sully, Ahmed, & Biddlecom, 2020). To our knowledge our study is the first that relies on large nationally representative survey data from three low- and middle-income countries to estimate the possible effects of COVID-19 on children’s experiences of violent discipline. 4.4 Implications Given the high levels of violence against children even prior to the pandemic and the potential impact of COVID-19 measures on risk factors associated with such violence, efforts to prevent and respond to violence against children should, be integrated as essential components of pandemic response and recovery plans (Bhatia et al., 2020). Although some measures enacted to contain COVID-19, such as school closures, may have reduced children’s exposure to specific forms of violence (e.g. school-based violence), children’s risk of violence in other settings including their homes and online remains high (Babvey et al., 2020). Increases in violent discipline in our models were driven mainly by large declines in happiness during periods of high COVID-19 restrictions. Wealth and changes to children’s time use patterns were comparatively less important, although these findings should be interpreted with caution as there are limitations in how wealth/income and children’s time use could be operationalized using MICS data. This important finding supports prioritization of mental health support for caregivers and families as a powerful way to mitigate the impact of the pandemic and reduce children’s exposure to violence in the home. Recognizing that times of hardship can also provide a window of opportunity to foster stronger relationships in the family, offering parents and caregivers guidance to build positive relationships and to manage conflict and stress should be a central component of strategies to prevent violence against children during the pandemic and beyond (Cluver et al., 2020). Measures to address families’ immediate needs, including paid sick leave for caregivers and child feeding programs, as well as longer-term social protection policies that reduce social inequities are equally fundamental to the pandemic response. Finally, a concerted effort is needed to improve the availability of quality population-level and administrative data on violence against children (Cappa & Petrowski, 2020). Under most circumstances, it is not advisable to collect data on direct experiences of violence within pandemic conditions for both ethical and methodological reasons. As restrictions lift, it will be important to invest in rigorous data collection to understand the impact of COVID-19 on the levels of violence, including testing model predictions, and to inform prevention and response strategies. 5 Conclusion Violence in all its forms represents an egregious violation of children’s right to a safe and healthy life. In the absence of robust population-level data on violence during COVID-19, governments and other agencies need to rely on alternative sources of evidence to formulate their prevention and response efforts. To help inform policy, we explored the possibilities and limits of using a multivariable predictive regression modelling approach to quantify changes in violent discipline under two different pandemic scenarios. We provide an approach which is accessible and can be used to predict changes in levels of violence under various pandemic scenarios, using robust national datasets and data on COVID-19’s impacts. This framework could be adapted for use with other datasets, in other countries, and assumptions updated as new data on the impacts of COVID-19 become available. Governments should plan for substantial increases in violent discipline under successive waves of ‘lockdown’ restrictions. Funding This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors. AB’s time was funded by an anonymous donor. Declaration of Competing Interest None. Appendix A Supplementary data The following is Supplementary data to this article: Appendix A Supplementary material related to this article can be found, in the online version, at doi:https://doi.org/10.1016/j.chiabu.2020.104897. ==== Refs References Babvey P. Capela F. Cappa C. Lipizzi C. Petrowski N. Ramirez-Marquez J. Using social media data for assessing children’s exposure to violence during the COVID-19 pandemic Child Abuse & Neglect 2020 104747 10.1016/j.chiabu.2020.104747 33358281 Bakrania S. Chávez C. Ipince A. Rocca M. Oliver S. Stansfield C. …Subrahmanian R. Impacts of pandemics and epidemics on child protection: Lessons learned from a rapid review in the context of COVID-19 Innocenti working papers no. 2020-05 2020 UNICEF Office of Research - Innocenti Florence Baron E.J. Goldstein E.G. Wallace C.T. Suffering in silence: How COVID-19 school closures inhibit the reporting of child maltreatment Journal of Public Economics 190 2020 104258 10.1016/j.jpubeco.2020.104258 32863462 Berger L.M. Income, family characteristics, and physical violence toward children Child Abuse & Neglect 29 2 2005 107 133 10.1016/j.chiabu.2004.02.006 15734178 Berman G. Ethical considerations for evidence generation involving children on the COVID-19 pandemic UNICEF innocenti discussion papers n.2020-01 2020 UNICEF Office of Research - Innocenti Florence Bhatia A. Fabbri C. Cerna-Turoff I. Tanton C. Knight L. Turner E. …Devries K. COVID-19 response measures and violence against children Bulletin of the World Health Organization 2020 98 2020 583 Biddle N. Edwards B. Gray M. Sollis K. Alcohol consumption during the COVID-19 period: May 2020 2020 Centre for Social Research and Methods, Australian National University Brown S.M. Doom J.R. Lechuga-Peña S. Watamura S.E. Koppels T. Stress and parenting during the global COVID-19 pandemic Child Abuse & Neglect 2020 104699 10.1016/j.chiabu.2020.104699 32859394 Bullinger L. Raissian K. Feely M. Schneider W. The neglected ones: Time at home during COVID-19 and child maltreatment Available at SSRN: https://ssrn.com/abstract=3674064 or https://doi.org/ 2020 10.2139/ssrn.3674064 Cabrera-Hernández F. Padilla-Romo M. Hidden violence: How COVID-19 school closures reduced the reporting of child maltreatment Working papers 2020-02 2020 University of Tennessee, Department of Economics Cappa C. Petrowski N. Thirty years after the adoption of the Convention on the Rights of the Child: Progress and challenges in building statistical evidence on violence against children Child Abuse & Neglect 2020 104460 10.1016/j.chiabu.2020.104460 32173109 Chawanpaiboon S. Vogel J.P. Moller A.B. Lumbiganon P. Petzold M. Hogan D. Gülmezoglu A.M. Global, regional, and national estimates of levels of preterm birth in 2014: a systematic review and modelling analysis The Lancet Global Health 7 1 2019 e37 e46 10.1016/s2214-109x(18)30451-0 30389451 Chung G. Lanier P. Wong P.Y.J. Mediating effects of parental stress on harsh parenting and parent-child relationship during coronavirus (COVID-19) pandemic in Singapore Journal of Family Violence 2020 10.1007/s10896-020-00200-1 Clay J.M. Parker M.O. Alcohol use and misuse during the COVID-19 pandemic: a potential public health crisis? The Lancet Public Health 5 5 2020 e259 10.1016/S2468-2667(20)30088-8 32277874 Cluver L. Lachman J.M. Sherr L. Wessels I. Krug E. Rakotomalala S. McDonald K. Parenting in a time of COVID-19 The Lancet Volume 395 Issue 10231 2020 e64 Cuartas J. McCoy D.C. Rey-Guerra C. Britto P.R. Beatriz E. Salhi C. Early childhood exposure to non-violent discipline and physical and psychological aggression in low- and middle-income countries: National, regional, and global prevalence estimates Child Abuse & Neglect 92 2019 93 105 10.1016/j.chiabu.2019.03.021 30939376 Devries K. Knight L. Petzold M. Merrill K.G. Maxwell L. Williams A. Abrahams N. Who perpetrates violence against children? A systematic analysis of age-specific and sex-specific data BMJ Paediatrics Open 2 1 2018 e000180 10.1136/bmjpo-2017-000180 29637183 Fisher J. Languilaire J.-C. Lawthom R. Nieuwenhuis R. Petts R.J. Runswick-Cole K. …Yerkes M.A. Community, work, and family in times of COVID-19 Community, Work & Family 23 3 2020 247 252 10.1080/13668803.2020.1756568 Hale T. Angrist N. Cameron-Blake E. Hallas L. Kira B. Majumdar S. Webster S. Oxford COVID-19 government response tracker 2020 Blavatnik School of Government ILO ILO monitor: COVID-19 and the world of work Updated estimates and analysis third edition 2020 Retrieved from: https://www.ilo.org/global/topics/coronavirus/impacts-and-responses/WCMS_743146/lang--en/index.htm ILO Youth & COVID-19: Impacts on jobs, education, rights and mental well-being. Survey Report 2020 Retrieved from: 2020 International Labour Organization https://www.ilo.org/global/topics/youth-employment/publications/WCMS_753026/lang--en/index.htm Jia R. Ayling K. Chalder T. Massey A. Broadbent E. Coupland C. …Vedhara K. Mental health in the UK during the COVID-19 pandemic: Cross-sectional analyses from a community cohort study BMJ Open 10 9 2020 e040620 10.1136/bmjopen-2020-040620 Kola L. Global mental health and COVID-19 The Lancet Psychiatry 7 8 2020 655 657 10.1016/S2215-0366(20)30235-2 32502468 Lawson M. Piel M.H. Simon M. Child maltreatment during the COVID-19 pandemic: Consequences of parental job loss on psychological and physical abuse towards children Child Abuse & Neglect 2020 104709 10.1016/j.chiabu.2020.104709 32893003 Maternowska M.C. Fry D. The multi-country study on the drivers of violence affecting children: An overview Vulnerable Children and Youth Studies 13 Suppl. 1 2018 12 25 10.1080/17450128.2018.1476748 Meinck F. Cluver L.D. Boyes M.E. Household illness, poverty and physical and emotional child abuse victimisation: Findings from South Africa’s first prospective cohort study BMC Public Health 15 1 2015 444 10.1186/s12889-015-1792-4 25924818 Ministry of Social Affairs and Public Housing Suriname multiple indicator cluster survey 2018 Survey findings report 2019 Ministry of Social Affairs and Public Housing. Paramaribo, Suriname Moller A.B. Petzold M. Chou D. Say L. Early antenatal care visit: A systematic analysis of regional and global levels and trends of coverage from 1990 to 2013 The Lancet Global Health 5 10 2017 e977 e983 10.1016/s2214-109x(17)30325-x 28911763 National Bureau of Statistics, & UNICEF Multiple Indicator cluster survey 2016-17, survey findings report 2017 National Bureau of Statistics and United Nations Children’s Fund Abuja, Nigeria National Statistical Office Social Indicator sample Survey-2018, survey findings report 2019 National Statistical Office of Mongolia Ulaanbaatar, Mongolia Parkes J. Datzberger S. Howell C. Kasidi J. Kiwanuka T. Knight L. …Devries K. Young people, inequality and violence during the COVID-19 lockdown in Uganda SocArXiv 2020 10.31235/osf.io/2p6hx Peterman A. O’Donnell M. COVID-19 and violence against women and children: A second research round up. A second research round up CGD Notes 2020 Center for Global Development Peterman A. Potts A. O’Donnell M. Thompson K. Shah N. Oertelt-Prigione s. …van Gelder N. Pandemics and violence against women and children CGD working paper 528 2020 Center for Global Development Washington, DC Petrowski N. Cappa C. Pereira A. Mason H. Daban R.A. Violence against children during COVID-19 Assessing and understanding change in use of helplines Child Abuse & Neglect 2020 104757 10.1016/j.chiabu.2020.104757 33593632 Raissian K.M. Bullinger L.R. Money matters: Does the minimum wage affect child maltreatment rates? Children and Youth Services Review 72 C 2017 60 70 Retrieved from https://EconPapers.repec.org/RePEc:eee:cysrev:v:72:y:2017:i:c:p:60-70 Riley T. Sully E. Ahmed Z. Biddlecom A. Estimates of the potential impact of the COVID-19 pandemic on sexual and reproductive health in low- and middle-income countries International Perspectives on Sexual and Reproductive Health 46 2020 73 76 10.1363/46e9020 32343244 Salari N. Hosseinian-Far A. Jalali R. Vaisi-Raygani A. Rasoulpoor S. Mohammadi M. Khaledi-Paveh B. Prevalence of stress, anxiety, depression among the general population during the COVID-19 pandemic: a systematic review and meta-analysis Globalization and Health 16 1 2020 57 10.1186/s12992-020-00589-w 32631403 Sedlak A.J. Mettenburg J. Basena M. Petta I. McPherson K. Greene A. …Li S. Fourth national incidence study of child abuse and neglect (NIS–4): Report to congress 2010 U.S. Department of Health and Human Services, Administration for Children and Families Washington, DC Serafini G. Parmigiani B. Amerio A. Aguglia A. Sher L. Amore M. The psychological impact of COVID-19 on the mental health in the general population QJM: An International Journal of Medicine 113 8 2020 531 537 10.1093/qjmed/hcaa201 32569360 Sharma A. Borah S.B. Covid-19 and domestic violence: An indirect path to social and economic crisis Journal of Family Violence 2020 10.1007/s10896-020-00188-8 Sidpra J. Abomeli D. Hameed B. Baker J. Mankad K. Rise in the incidence of abusive head trauma during the COVID-19 pandemic Archives of Disease in Childhood 2020 10.1136/archdischild-2020-319872 archdischild-2020-319872 Stith S.M. Liu T. Davies L.C. Boykin E.L. Alder M.C. Harris J.M. Dees J.E.M.E.G. Risk factors in child maltreatment: A meta-analytic review of the literature Aggression and Violent Behavior 14 1 2009 13 29 10.1016/j.avb.2006.03.006 The Alliance for Child Protection in Humanitarian Action Technical note: Protection of children during the coronavirus pandemic, version 2 Retrieved from: 2020 https://alliancecpha.org/en/COVD19 The New York Times Covid world map: Tracking the global outbreak Updated September 7, 2020 2020 The World Bank COVID-19 crisis through a migration Lens Migration and development brief, no. 32 2020 World Bank Washington, DC The World Bank Global economic prospects Retrieved from: 2020 https://www.worldbank.org/en/publication/global-economic-prospects UNICEF Child disciplinary practices at home: Evidence from a range of low- and middle-income countries 2010 United Nations Children’s Fund New York UNICEF Hidden in Plain Sight: A statistical analysis of violence against children 2014 United Nations Children’s Fund New York UNICEF A Familiar Face: Violence in the lives of children and adolescent 2017 United Nations Children’s Fund New York UNICEF The State of the world’s children 2019 Children, Food and Nutrition: Growing well in a changing world 2019 United Nations Children’s Fund New York UNICEF Evidence Gap Map: Pandemics, epidemics and outcomes on child protection and violence Retrieved from: 2020 https://www.unicef-irc.org/evidence-gap-map-pandemics-child-protection-violence/ UNICEF Research on Violence against Children during the COVID-19 Pandemic: Guidance to inform ethical data collection and evidence generation 2020 United Nations Children’s Fund New York United Nations Convention on the rights of the child New York, 20 November 1989 1989 United Nations Policy brief: COVID-19 and the need for action on mental health UNITED NATIONS policy brief 2020 Whipple E.E. Webster-Stratton C. The role of parental stress in physically abusive families Child Abuse & Neglect 15 3 1991 279 291 10.1016/0145-2134(91)90072-L 2043979 WHO Addressing violence against children, women and older people during the COVID-19 pandemic: Key actions 17 June 2020. WHO/2019-nCoV/Violence_actions/2020.1 2020

33451678

PMC9754318

NO-CC CODE

2022-12-16 23:26:36

no

Child Abuse Negl. 2021 Jun 22; 116:104897

utf-8

Child Abuse Negl

10.1016/j.chiabu.2020.104897

oa_other